scripts/notebooks/Results.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MARS-Gym Experiments - Trivago Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "        <script type=\"text/javascript\">\n",
       "        window.PlotlyConfig = {MathJaxConfig: 'local'};\n",
       "        if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}\n",
       "        if (typeof require !== 'undefined') {\n",
       "        require.undef(\"plotly\");\n",
       "        requirejs.config({\n",
       "            paths: {\n",
       "                'plotly': ['https://cdn.plot.ly/plotly-latest.min']\n",
       "            }\n",
       "        });\n",
       "        require(['plotly'], function(Plotly) {\n",
       "            window._Plotly = Plotly;\n",
       "        });\n",
       "        }\n",
       "        </script>\n",
       "        "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/media/workspace/DeepFood/deep-reco-gym/src/mars_gym/tools/eval_viz/app.py:13: FutureWarning:\n",
      "\n",
      "Passing a negative integer is deprecated in version 1.0 and will not be supported in future version. Instead, use None to not limit the column width.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "import sys, os\n",
    "os.environ['PATH_EVALUATION'] =  \"../../output/evaluation/\"\n",
    "os.environ['PATH_EVAL_REINFORCEMENT'] =  \"../../output/interaction/\"\n",
    "os.environ['PATH_TRAIN'] =  \"../../output/models/\"\n",
    "\n",
    "#import plotly.graph_objects as go\n",
    "import plotly.graph_objects as go\n",
    "from plotly.offline import init_notebook_mode, iplot\n",
    "init_notebook_mode(connected=True) \n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import json\n",
    "from pandas.io.json import json_normalize\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from mars_gym.tools.eval_viz.app import fetch_iteraction_results_path, load_all_iteraction_metrics, fetch_results_path, load_data_metrics, load_eval_params, filter_df, load_fairness_metrics, load_fairness_df\n",
    "from mars_gym.tools.eval_viz.plot import plot_line_iteraction, plot_exploration_arm, get_colors, confidence, plot_bar\n",
    "from  mars_gym.tools.eval_viz.app import PATH_EVAL_REINFORCEMENT\n",
    "\n",
    "from util import list_paths_per_model, load_iteractions_params, get_metrics_reward, group_metrics, load_dataset, plot_cum_reward, plot_fairness_metrics, plot_fairness_mistreatment, plot_fairness_treatment\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook includes all viz of paper **'MARS-Gym: A Gym framework to model, train, and evaluate recommendation systems for marketplaces'**  and can be used for reproducibility or example of framework usage. \n",
    "\n",
    "We need run some scripts before vizualise results:\n",
    "* scripts/simulation/chicago_usa_script.sh\n",
    "* scripts/metrics/metrics_chicago_usa_script.sh\n",
    "* scripts/metrics/fairness_recsys_script.sh"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__a06cfb73a9',\n",
       " '../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__f9e2903219',\n",
       " '../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237',\n",
       " '../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___explore_rounds_0b659e17e7',\n",
       " '../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___epsilon___0_1__76d6a2cc36']"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result_paths = list_paths_per_model(PATH_EVAL_REINFORCEMENT)\n",
    "result_paths[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/media/workspace/DeepFood/mars-gym-experiments/scripts/notebooks/util.py:62: FutureWarning:\n",
      "\n",
      "pandas.io.json.json_normalize is deprecated, use pandas.json_normalize instead\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>project</th>\n",
       "      <th>minimum_interactions</th>\n",
       "      <th>session_test_size</th>\n",
       "      <th>dataset_split_method</th>\n",
       "      <th>val_size</th>\n",
       "      <th>n_splits</th>\n",
       "      <th>split_index</th>\n",
       "      <th>sampling_strategy</th>\n",
       "      <th>balance_fields</th>\n",
       "      <th>use_sampling_in_validation</th>\n",
       "      <th>...</th>\n",
       "      <th>bandit_policy_params.alpha</th>\n",
       "      <th>iteraction</th>\n",
       "      <th>bandit_policy_params.exploration_threshold</th>\n",
       "      <th>bandit_policy_params.decay_rate</th>\n",
       "      <th>bandit_policy_params.explore_rounds</th>\n",
       "      <th>bandit_policy_params.epsilon</th>\n",
       "      <th>loss_function_params.clip</th>\n",
       "      <th>bandit_policy_params.logit_multiplier</th>\n",
       "      <th>bandit_policy_params.arg</th>\n",
       "      <th>bandit_policy_params.v_sq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trivago.config.trivago_experiment</td>\n",
       "      <td>5</td>\n",
       "      <td>0.1</td>\n",
       "      <td>time</td>\n",
       "      <td>0.2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>none</td>\n",
       "      <td>[]</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00001</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__a06cfb73a9</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trivago.config.trivago_experiment</td>\n",
       "      <td>5</td>\n",
       "      <td>0.1</td>\n",
       "      <td>time</td>\n",
       "      <td>0.2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>none</td>\n",
       "      <td>[]</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00001</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__f9e2903219</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trivago.config.trivago_experiment</td>\n",
       "      <td>5</td>\n",
       "      <td>0.1</td>\n",
       "      <td>time</td>\n",
       "      <td>0.2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>none</td>\n",
       "      <td>[]</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237</td>\n",
       "      <td>0.7</td>\n",
       "      <td>0.000097</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trivago.config.trivago_experiment</td>\n",
       "      <td>5</td>\n",
       "      <td>0.1</td>\n",
       "      <td>time</td>\n",
       "      <td>0.2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>none</td>\n",
       "      <td>[]</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___explore_rounds_0b659e17e7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000187</td>\n",
       "      <td>1000.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>trivago.config.trivago_experiment</td>\n",
       "      <td>5</td>\n",
       "      <td>0.1</td>\n",
       "      <td>time</td>\n",
       "      <td>0.2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>none</td>\n",
       "      <td>[]</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___epsilon___0_1__76d6a2cc36</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 55 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                             project  minimum_interactions  session_test_size  \\\n",
       "0  trivago.config.trivago_experiment  5                     0.1                 \n",
       "0  trivago.config.trivago_experiment  5                     0.1                 \n",
       "0  trivago.config.trivago_experiment  5                     0.1                 \n",
       "0  trivago.config.trivago_experiment  5                     0.1                 \n",
       "0  trivago.config.trivago_experiment  5                     0.1                 \n",
       "\n",
       "  dataset_split_method  val_size  n_splits  split_index sampling_strategy  \\\n",
       "0  time                 0.2       5         0            none               \n",
       "0  time                 0.2       5         0            none               \n",
       "0  time                 0.2       5         0            none               \n",
       "0  time                 0.2       5         0            none               \n",
       "0  time                 0.2       5         0            none               \n",
       "\n",
       "  balance_fields  use_sampling_in_validation  ...  bandit_policy_params.alpha  \\\n",
       "0  []             False                       ...  0.00001                      \n",
       "0  []             False                       ...  0.00001                      \n",
       "0  []             False                       ... NaN                           \n",
       "0  []             False                       ... NaN                           \n",
       "0  []             False                       ... NaN                           \n",
       "\n",
       "                                                                                                                 iteraction  \\\n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__a06cfb73a9   \n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__f9e2903219   \n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237   \n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___explore_rounds_0b659e17e7   \n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___epsilon___0_1__76d6a2cc36   \n",
       "\n",
       "   bandit_policy_params.exploration_threshold bandit_policy_params.decay_rate  \\\n",
       "0 NaN                                         NaN                               \n",
       "0 NaN                                         NaN                               \n",
       "0  0.7                                         0.000097                         \n",
       "0 NaN                                          0.000187                         \n",
       "0 NaN                                         NaN                               \n",
       "\n",
       "  bandit_policy_params.explore_rounds  bandit_policy_params.epsilon  \\\n",
       "0 NaN                                 NaN                             \n",
       "0 NaN                                 NaN                             \n",
       "0 NaN                                 NaN                             \n",
       "0  1000.0                             NaN                             \n",
       "0 NaN                                  0.1                            \n",
       "\n",
       "   loss_function_params.clip bandit_policy_params.logit_multiplier  \\\n",
       "0 NaN                        NaN                                     \n",
       "0 NaN                        NaN                                     \n",
       "0 NaN                        NaN                                     \n",
       "0 NaN                        NaN                                     \n",
       "0 NaN                        NaN                                     \n",
       "\n",
       "   bandit_policy_params.arg  bandit_policy_params.v_sq  \n",
       "0 NaN                       NaN                         \n",
       "0 NaN                       NaN                         \n",
       "0 NaN                       NaN                         \n",
       "0 NaN                       NaN                         \n",
       "0 NaN                       NaN                         \n",
       "\n",
       "[5 rows x 55 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We will only filter the city of Chicago to exemplify graphs and results.\n",
    "#\n",
    "\n",
    "#city = \"Rio de Janeiro, Brazil\"\n",
    "#city = \"Como, Italy\"\n",
    "city = \"Chicago, USA\"\n",
    "#city = \"New York, USA\"\n",
    "#city = 'Copenhagen, Denmark'\n",
    "#city='recsys'\n",
    "\n",
    "df_params = load_iteractions_params(result_paths)\n",
    "df_params = df_params[df_params['data_frames_preparation_extra_params.filter_city'] == city]\n",
    "df_params.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>bandit_policy_class</th>\n",
       "      <th>observation</th>\n",
       "      <th>iteraction</th>\n",
       "      <th>name</th>\n",
       "      <th>model</th>\n",
       "      <th>model_list</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "      <td></td>\n",
       "      <td>[../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_83ef3f55d3, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_42716ae133, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_520cebc570, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d70af8a008, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6b44643887, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4]</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc</td>\n",
       "      <td>[../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_83ef3f55d3, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_42716ae133, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_520cebc570, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d70af8a008, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6b44643887, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>mars_gym.model.bandit.CustomRewardModelLinUCB</td>\n",
       "      <td></td>\n",
       "      <td>[../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__ef475653f1, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__75eccc229a, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__bad81a6465, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__debd700fb8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__2165771e8e, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__0b623b96b8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__d5981787b1]</td>\n",
       "      <td>mars_gym.model.bandit.CustomRewardModelLinUCB</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b</td>\n",
       "      <td>[../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__ef475653f1, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__75eccc229a, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__bad81a6465, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__debd700fb8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__2165771e8e, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__0b623b96b8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__d5981787b1]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             bandit_policy_class observation  \\\n",
       "0  mars_gym.model.bandit.AdaptiveGreedy                        \n",
       "1  mars_gym.model.bandit.CustomRewardModelLinUCB               \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         iteraction  \\\n",
       "0  [../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_83ef3f55d3, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_42716ae133, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_520cebc570, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d70af8a008, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6b44643887, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4]   \n",
       "1  [../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__ef475653f1, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__75eccc229a, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__bad81a6465, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__debd700fb8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__2165771e8e, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__0b623b96b8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__d5981787b1]   \n",
       "\n",
       "                                            name  \\\n",
       "0  mars_gym.model.bandit.AdaptiveGreedy            \n",
       "1  mars_gym.model.bandit.CustomRewardModelLinUCB   \n",
       "\n",
       "                                                                                                                      model  \\\n",
       "0  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc   \n",
       "1  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         model_list  \n",
       "0  [../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6822e93237, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_83ef3f55d3, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_42716ae133, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_520cebc570, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d70af8a008, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_6b44643887, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4]  \n",
       "1  [../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__ef475653f1, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__75eccc229a, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__bad81a6465, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__debd700fb8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__2165771e8e, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__0b623b96b8, ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__d5981787b1]  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Group results\n",
    "\n",
    "list_bandits          = df_params.groupby([\"bandit_policy_class\", \"observation\"])['iteraction'].apply(list).reset_index()\n",
    "list_bandits['name']  = list_bandits.bandit_policy_class.map(str) + ' ' + list_bandits.observation.map(str)\n",
    "list_bandits['name']  = list_bandits['name'].apply(lambda x: x.replace(\"fixed \", \"\").replace(\" \", \"\"))\n",
    "list_bandits['model'] = list_bandits.iteraction.apply(lambda l: l[np.random.randint(0, len(l))])\n",
    "list_bandits['model_list'] = list_bandits.iteraction\n",
    "list_bandits.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "reward_metrics = []\n",
    "for i, row in list_bandits.iterrows():\n",
    "    df_metrics = get_metrics_reward(row['iteraction'])\n",
    "    df_metrics['bandit'] = row['name']\n",
    "    df_metrics['model']  = row['model']\n",
    "\n",
    "    df_g_metrics = group_metrics(df_metrics)\n",
    "    reward_metrics.append(df_g_metrics)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>r_mean|mean</th>\n",
       "      <th>r_mean|std</th>\n",
       "      <th>r_reward|mean</th>\n",
       "      <th>r_reward|std</th>\n",
       "      <th>r_reward|count</th>\n",
       "      <th>model|first</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bandit</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.SoftmaxExplorer</th>\n",
       "      <td>0.275622</td>\n",
       "      <td>0.013914</td>\n",
       "      <td>2915.250000</td>\n",
       "      <td>147.173124</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___logit_multipli_34358aae37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.RandomPolicy</th>\n",
       "      <td>0.040749</td>\n",
       "      <td>NaN</td>\n",
       "      <td>431.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b____1649c7bd6b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.PercentileAdaptiveGreedy</th>\n",
       "      <td>0.284485</td>\n",
       "      <td>0.010725</td>\n",
       "      <td>3009.000000</td>\n",
       "      <td>113.439726</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_36a2721c50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.LinUCB</th>\n",
       "      <td>0.121703</td>\n",
       "      <td>0.007030</td>\n",
       "      <td>1287.250000</td>\n",
       "      <td>74.351962</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__335eb27eb5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.LinThompsonSampling</th>\n",
       "      <td>0.048501</td>\n",
       "      <td>NaN</td>\n",
       "      <td>513.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___v_sq___0_1__4cd65ca01d</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.FixedPolicyPopularItem</th>\n",
       "      <td>0.104567</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1106.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___arg___2__9cb6645207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.ExploreThenExploit</th>\n",
       "      <td>0.228906</td>\n",
       "      <td>0.006917</td>\n",
       "      <td>2421.142857</td>\n",
       "      <td>73.156063</td>\n",
       "      <td>7</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___explore_rounds_a180ca8b8f</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.EpsilonGreedy</th>\n",
       "      <td>0.285147</td>\n",
       "      <td>0.015851</td>\n",
       "      <td>3016.000000</td>\n",
       "      <td>167.654406</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___epsilon___0_1__4645466ef3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.CustomRewardModelLinUCB</th>\n",
       "      <td>0.297863</td>\n",
       "      <td>0.024824</td>\n",
       "      <td>3150.500000</td>\n",
       "      <td>262.564822</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.AdaptiveGreedy</th>\n",
       "      <td>0.215609</td>\n",
       "      <td>0.030293</td>\n",
       "      <td>2280.500000</td>\n",
       "      <td>320.404209</td>\n",
       "      <td>8</td>\n",
       "      <td>../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                r_mean|mean  r_mean|std  \\\n",
       "bandit                                                                    \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.275622     0.013914     \n",
       "mars_gym.model.bandit.RandomPolicy              0.040749    NaN           \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.284485     0.010725     \n",
       "mars_gym.model.bandit.LinUCB                    0.121703     0.007030     \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.048501    NaN           \n",
       "mars_gym.model.bandit.FixedPolicyPopularItem    0.104567     0.000000     \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.228906     0.006917     \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.285147     0.015851     \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.297863     0.024824     \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.215609     0.030293     \n",
       "\n",
       "                                                r_reward|mean  r_reward|std  \\\n",
       "bandit                                                                        \n",
       "mars_gym.model.bandit.SoftmaxExplorer           2915.250000    147.173124     \n",
       "mars_gym.model.bandit.RandomPolicy              431.000000    NaN             \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  3009.000000    113.439726     \n",
       "mars_gym.model.bandit.LinUCB                    1287.250000    74.351962      \n",
       "mars_gym.model.bandit.LinThompsonSampling       513.000000    NaN             \n",
       "mars_gym.model.bandit.FixedPolicyPopularItem    1106.000000    0.000000       \n",
       "mars_gym.model.bandit.ExploreThenExploit        2421.142857    73.156063      \n",
       "mars_gym.model.bandit.EpsilonGreedy             3016.000000    167.654406     \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   3150.500000    262.564822     \n",
       "mars_gym.model.bandit.AdaptiveGreedy            2280.500000    320.404209     \n",
       "\n",
       "                                                r_reward|count  \\\n",
       "bandit                                                           \n",
       "mars_gym.model.bandit.SoftmaxExplorer           8                \n",
       "mars_gym.model.bandit.RandomPolicy              1                \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  8                \n",
       "mars_gym.model.bandit.LinUCB                    8                \n",
       "mars_gym.model.bandit.LinThompsonSampling       1                \n",
       "mars_gym.model.bandit.FixedPolicyPopularItem    2                \n",
       "mars_gym.model.bandit.ExploreThenExploit        7                \n",
       "mars_gym.model.bandit.EpsilonGreedy             8                \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   8                \n",
       "mars_gym.model.bandit.AdaptiveGreedy            8                \n",
       "\n",
       "                                                                                                                                                             model|first  \n",
       "bandit                                                                                                                                                                    \n",
       "mars_gym.model.bandit.SoftmaxExplorer           ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___logit_multipli_34358aae37  \n",
       "mars_gym.model.bandit.RandomPolicy              ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b____1649c7bd6b                \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_36a2721c50  \n",
       "mars_gym.model.bandit.LinUCB                    ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__335eb27eb5  \n",
       "mars_gym.model.bandit.LinThompsonSampling       ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___v_sq___0_1__4cd65ca01d     \n",
       "mars_gym.model.bandit.FixedPolicyPopularItem    ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___arg___2__9cb6645207        \n",
       "mars_gym.model.bandit.ExploreThenExploit        ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___explore_rounds_a180ca8b8f  \n",
       "mars_gym.model.bandit.EpsilonGreedy             ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___epsilon___0_1__4645466ef3  \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___alpha___1e_05__7966e4e52b  \n",
       "mars_gym.model.bandit.AdaptiveGreedy            ../../output/interaction/InteractionTraining/results/InteractionTraining____mars_gym_model_b___exploration_th_d3663f58fc  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_all = pd.concat(reward_metrics)\n",
    "df_all = df_all.reset_index().sort_values('bandit', ascending=False).set_index('bandit')\n",
    "df_all"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DataViz - Plot Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['mars_gym.model.bandit.AdaptiveGreedy',\n",
       "       'mars_gym.model.bandit.CustomRewardModelLinUCB',\n",
       "       'mars_gym.model.bandit.EpsilonGreedy',\n",
       "       'mars_gym.model.bandit.ExploreThenExploit',\n",
       "       'mars_gym.model.bandit.FixedPolicyPopularItem',\n",
       "       'mars_gym.model.bandit.LinThompsonSampling',\n",
       "       'mars_gym.model.bandit.LinUCB',\n",
       "       'mars_gym.model.bandit.PercentileAdaptiveGreedy',\n",
       "       'mars_gym.model.bandit.RandomPolicy',\n",
       "       'mars_gym.model.bandit.SoftmaxExplorer'],\n",
       "      dtype='object', name='name')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#list_bandits['name'] = list_bandits['name'].apply(lambda x: \"popular_item\" if x == \"PopularItem\" else x)\n",
    "df_models = list_bandits.set_index('name')[['model_list']]\n",
    "df_models.index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>iteraction</th>\n",
       "      <th>idx</th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>reward</th>\n",
       "      <th>ps</th>\n",
       "      <th>index_env</th>\n",
       "      <th>_idx</th>\n",
       "      <th>mean_reward</th>\n",
       "      <th>bandit</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4</td>\n",
       "      <td>0</td>\n",
       "      <td>559</td>\n",
       "      <td>191</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.975610</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4</td>\n",
       "      <td>1</td>\n",
       "      <td>1118</td>\n",
       "      <td>486</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.951220</td>\n",
       "      <td>43</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4</td>\n",
       "      <td>2</td>\n",
       "      <td>1677</td>\n",
       "      <td>644</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.926829</td>\n",
       "      <td>96</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4</td>\n",
       "      <td>3</td>\n",
       "      <td>1908</td>\n",
       "      <td>1263</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.902439</td>\n",
       "      <td>186</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4</td>\n",
       "      <td>4</td>\n",
       "      <td>1971</td>\n",
       "      <td>1397</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.878049</td>\n",
       "      <td>297</td>\n",
       "      <td>5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>mars_gym.model.bandit.AdaptiveGreedy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                            iteraction  idx  \\\n",
       "0  InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4  0     \n",
       "1  InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4  1     \n",
       "2  InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4  2     \n",
       "3  InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4  3     \n",
       "4  InteractionTraining____mars_gym_model_b___exploration_th_2cc29714f4  4     \n",
       "\n",
       "   user  item  reward        ps  index_env  _idx  mean_reward  \\\n",
       "0  559   191   0.0     0.975610  10         1     0.0           \n",
       "1  1118  486   0.0     1.951220  43         2     0.0           \n",
       "2  1677  644   0.0     2.926829  96         3     0.0           \n",
       "3  1908  1263  0.0     3.902439  186        4     0.0           \n",
       "4  1971  1397  0.0     4.878049  297        5     0.0           \n",
       "\n",
       "                                 bandit  \n",
       "0  mars_gym.model.bandit.AdaptiveGreedy  \n",
       "1  mars_gym.model.bandit.AdaptiveGreedy  \n",
       "2  mars_gym.model.bandit.AdaptiveGreedy  \n",
       "3  mars_gym.model.bandit.AdaptiveGreedy  \n",
       "4  mars_gym.model.bandit.AdaptiveGreedy  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_size       = 15000\n",
    "window_size       = 1000 \n",
    "bandits           = df_models.index\n",
    "\n",
    "df_group = load_dataset(df_models, bandits, sample_size)\n",
    "df_group.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Bandit Simulation Results."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We performed the simulation of each proposed task until the convergence of most\n",
    "methods to observe cumulative mean reward throughout the simulation. Bandit Simulation Results for \"Chicago, USA\" task."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAw4AAAESCAYAAABUyYKEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeVgU5QMH8O+eLDe7y32JIrzgCeKZZ6DiLaZpmpkdppb4i8yiVFJSQ80bxas8srwvNA0Vr7SyVDzBQZD7ZtldYJdlr/n9wREa6mqe+X6eh8fdmXfeY2aQec/hsCwLiqIoiqIoiqKo++E+6wxQFEVRFEVRFPX8oxUHiqIoiqIoiqIeiFYcKIqiKIqiKIp6IFpxoCiKoiiKoijqgWjFgaIoiqIoiqKoB6IVB4qiKIqiKIqiHoj/tBMkhPgC2AxACkAGYBzDMLfuEZYASAKwmmGYT2u3WQDYCCAIgB7ApwzDHHoaeacoiqIoiqKol9Wz6HFYA2AVwzC+AFYBWNtYIEIIr3bf/rt2fQqgnGGY5gAGA9hACLF6gvmlKIqiKIqiqJfeU604EEIcAbQDsK120zYA7QghDo0EjwRwCEDqXdtHobayUdtTcQFA/yeSYYqiKIqiKIqiADz9oUoeAPIYhjEAAMMwBkJIfu32krpAhJC2AEIBvApg1l1xeALIavA9u/Z4U5gB6ACgAIDhUQpAURT1EuIBcAHwF4DqZ5wXiqIo6hl56nMcHoQQIgCwDsA7tRWLxxl9BwC/Ps4IKYqiXiLdAZx91IMvXrzoyOXyNgBoBbo4B0VR1POE5XA4SqPRsJFl2bigoCBtY4GedsUhB4AbIYRXWyngAXCt3V7HBYA3gMO1lQY7ABxCiA3DMB+gpoehCf7uofAEcNLE9AsAQC5XwWhkHzrzUqkVZLLKhz7uRfMylPNlKCNAy/lf8izLyOVyIBZbArX/hz56PLwNNjYSf2trOzmHw3n4/4QpiqKoJ4JlWRgMeoFCURqu0ajbAXi7sXBPteLAMEwxIeQygNEAttb+m8QwTEmDMNkA7Ou+E0JmA7CqW1UJwC4AEwFcIIT4oKYXYbSJWTAAgNHIPlLFoe7Yl8HLUM6XoYwALed/yXNQxn87xLMVrTRQFEU9fzgcDvh8gU4icVLm52d0u1e4Z9FVPAlAOCEkFUB47XcQQg4TQtqbcPwiAHaEkDTUTJ7+gGGYiieWW4qiKOpx4dJKA0VR1POLy+WyqJnX1qinPseBYZibADo1sn3APcLPvuu7CsDrTyRzFEVRFEVRFEU1ik5OoyiKol56YWH9Wp85c8r6aaYTF7fSOSoqssmTTpOiKOpxee5WVaIoiqKol8HkyeGFdZ+zszOFY8aMaH3q1B8X+Xz6p5miqOcT7XGgKIqiKIqiKOqBaLMGRVEURQFITr5uGRu71FOhkAs6duysmDkzOquqqoobFRXZNC0t1dJgMHAI8a+MjJyV5ebmrgOACRPGkZYt21RcuZJkk5OTZe7jQyrnzl2QIZXa6wFg796dks2bv3fTaDS8oUNfK2yY3ooVS1zz8/PMYmIWZ0yZMpEAQGhoz0AAWLBgaWr79h1VT/scUBRF3Q/tcaAoiqIoACdPHpcuWRKbumPH/mt5eXmiuLiVLixrRP/+g0r37Pn52p49P18VCoXGRYvmezY87vTpE5IZM2ZnHDx47LJer+Nu3LjBCQAYJkW0atXyJpGRszLi4xOulJcr+WVlZcLG0o6NXcsAQELC6aTExHNJtNJAUdTziFYcKIqiKArAkCHDit3dPXRiscQwduz4gjNnTkokEqlhwIDBCgsLC6O1tbVx/PgJBcnJ1++YRN27d6iseXOfanNzc7ZHj+Cy27fTLADg2LEEcbt2HZRdunStNDMzY8PDP8mny9FSFPUio0OVKIqiKAqAk5OLtu6zm5tbtUIhF6rVau7ChfM8kpIu2qhUlXwA0Gg0XIPBAB6vZqlzqdReV3ecSCQyajQaLgDIZCUCBweH+jgtLS2NVlbW+qdWIIqiqMeMVhwoiqIoCkBRUUH9MKL8/DyhnZ1Yu3HjOqe8vBzRunWbUpycnPXXrl01nzz53RYs++COA4nEXpednWVe912tVnMrKysa/bvL4XAeSxkoiqKeJDpUiaIoiqIAHDy43yEvL1cgl5fxtm7d7NK9e0+5Wq3mCYVmRltbO4NcXsb77rs1rqbG16dPqPzSpb9sz5//3Uqr1XJWrlzqyrJsozUEqdRBz+FwkJmZYfb4SkRRFPV43bPHgRDS42EiYhjmzL/PDkVRFEU9G716hZRFRHzkK5fLBR06dFJMmhReoFQqeFFRkc0GDgwJsLMT64YPH1l44cKfdqbE5+fXQjN58tTsefNmN62uruYNHfpaoUQi0TYW1sLCwjhixKiC8PAP/PR6Aycm5ttbQUF0gjRFUc8Xzr26WwkhRgAsgLrWkYYBOXd9B8MwvCeRwcfMC0CGTFYJo/Hh56c5OFijpKTisWfqefMylPNlKCNAy/lf8izLyOVyIJVaAUBTAJmPGk9S0uVMd3fv0seVL4qiKOrxy81Ntw8MDPBqbN/95ji0bvDZBcD3AH4BsBdAMQBHAMMBhAJ497HklKIoiqIoiqKo59I9Kw4Mw9yo+0wImQ9gC8MwM+8K9gshZC6AjwEcfzJZpCiKoiiKoijqWTN1cnQIgNP32HcaQK/HkhuKoiiKoiiKop5LplYcygAMvce+YbX7KYqiKIqiKIr6jzL1PQ4xAGIJIV4A4vH3HIehAPoDmPJEckdRFEVRFEVR1HPBpIoDwzCrCSF5AL4EsAoAD4ABQBKA1xiG2f/kskhRFEVRFEVR1LP2wIoDIUQAoCOAvxiG6UQI4QGwB1DKMIzhSWeQoiiKoiiKoqhnz5QeBwOAE6gZkpRfW1koetQECSG+ADYDkAKQARjHMMytu8K8AyACgBE1vRvrGYZZUbtvNoAPAeTXBj/HMMxHj5ofiqIoiqIoiqIe7IGToxmGMQK4BcD5MaW5BsAqhmF8UTPsaW0jYfYAaMswTACAVwBMI4S0abB/C8MwAbU/tNJAURRFUc+5CRPGke3bf7Q3JWy3bu2Dbt9ON2tsX1hYv9Znzpyyfry5a1zDtOLiVjpHRUU2eVJp7dq1Xfruu2PJk4g7NzdbGBLSNdBgePEHivz226/Wgwf3bfPgkNSTYOqqSjMARBFCWj8w5H0QQhwBtAOwrXbTNgDtCCEODcMxDFPOMEzdq50tAAhw15uqKYqiKIqinpbJk8MLo6NjsgAgOztT2K1b+yC9Xv/A41asWOLarVv7oIsX/7J84pmsdXflyt3dU5uYeC6Jx+P967grKiq4MTFfu4eF9WsdHNw1cMiQ0NaffBLe7GmWj3p2TF1VaSZqhhZdrp0kXYS7HuQZhuloQjweAPLq5kYwDGMghOTXbi9pGJAQMgTANwC8AXzBMMy1BrvfIIT0BVAI4CuGYX43sRwURVEU9dQYDAY8joc16sXEsixOnjwutbS0Mhw+fFAaFNRB9azz9G9UV1dzPvzwfWJpaan/5pvFt3x9/TQajYZ76lSi7blzv9o2Vj69Xg8+39THTep5Z+qVvF7789QwDBMPIJ4Q4glgPyHkMMMwDGqGOs1jGEZHCOkD4AAhxJ9hGJmpcUulVo+cLweHp9I7+sy9DOV8GcoI0HL+l7wMZXxWwsL6tR44cGjxiRPHpEVFRWZdu3YvCw+PyJszZ2ZThkmx8vb2UcXELEkXi8WGadPCm6WkJFvrdFqOp6dX1fTpX2T5+bXQAMCMGdO9hEIzY3FxkTAlJdl69ux5aTqdjrt2bax7WZlMKBKZG8LChhe9//6k+84VXLdutdOBA3udOBxg7Nh38mNjlzbZsmXHdaVSwZsxY7rPwYPHrtRVSI4cOWS3ZctG123b9iSvWLHENSsrQyQQCNkLF87b2ds7VM+duzD9+PEE8YEDe50EAj47bVpkZs+eweWNpdutW/ugiROnZO/fv9tJqVQIBg0KKxo69DVZdPTMpjk52eYBAe2U8+d/myEUClkA2LZtq/2uXducVapKPiH+lV98EZXl4uKqA4DTp0/axMYu9VAo5IKePYPL2LvGDezc+ZN0167tzkqlQuDt7aP68suoLA+PJlpTrldy8nXL2NilngqFXNCxY2fFzJnRWSKRiJXL5byoqMimaWmplgaDgUOIf2Vk5KwsNzd3HVAzXKplyzYVV64k2eTkZJn7+JDKuXMXZEil9noA2Lt3p2Tz5u/dNBoNb+jQ1wobprlixRLX/Pw8s5iYxRlTpkwkABAa2jMQABYsWJravn3Hfzw0//HHb1YKhVwwdeq0zDVrYj21Wm1O3bkrK5PxZs+e0TQ5+bq1q6ubpl27DsqGx86fP8fjjz/OidVqNc/Z2UUzZUpETufOr1TW5SUzM0PE43GRlHTR1snJRfPFF1GZrVq1rvrii2lNZTKZ8KuvvvDhcrnsqFFv5vfrN0A+ZsyI1qdO/XHxyJGD4p07tzn/8MPOlLq0vv9+nePly5dsVqxYk1ZdXc1Ztuxbt3Pnzoj1eh23U6cu8s8+m5ljbm7O7tu3W1pWJhOsWfM9Y2lpaQQAS0tL48CBQ+QA5HfeRx9l79+/x8lgMHD27z9y7fjxBNvvv1/nVlJSInR3d9d8+umXWS1btqoCgIKCfMGiRfM9U1JuWJmZiYxhYcOLxo9/vxgAqqqqOHPnRjX566/zdnZ2Yl2fPv1K69JZvz7OKSXlhtWSJbHpddvmzZvtweFw8OWXX+WYci9RD8fU5VjfeUzp5QBwI4TwansbeABca7ffK+1sQsifAAbVfGUKG+w7RgjJAdAK936z9T/IZJUwGh9+5JODgzVKSioe+rgXzctQzpehjAAt53/Jsywjl8v5Vw0ujUm8mCv97XqBSePdH9UrrVxKQ4LcTW5UOnv2jHj58rhUvV7Pee+9t1pEREyxiIycmenjQzRTp07y2bp1k2N4eERBp05dlLNnz88UCoXs4sUL3KOjZzX76ac9yX/Hc1oyb96iW+3bd0zTarWc4cMHtZ41K/p2586vVMrlcl5OTlajY/frnDx53Gb//j3OS5asZDw9vbTR0bPqx9UHBgaprays9adPn7AJDu5TDgAJCYelISF96st58eIFu9mz56bNnbsgY9asSK9PP53q27dv/5L4+ISru3fvkC5duqhJz57B1xpLGwAuXDhvs3HjT8n5+bnCSZPebZGSct3qq6/m3RaLJYYJE8b5xcfvk4wYMUp29uwZ682bN7gtXLjsFiH+VYsWzXefOfPzZt999wMjk5Xyv/56lndExGeZffv2V/zww0aHo0ePOPTuHSqrzbPd9u0/usTELE5r1qy5Zu3aVS5RUV8027jxp5umXKuTJ49LlyyJTbW0tDR+8km4T1zcSpeIiOn5LGtE//6DSnv1CrltMBgwe/aXXosWzfdctmx1/YPl6dMnJIsWLb/l5uaunTp1ou/GjRucPv00Mo9hUkSrVi1vMnfuwlvt2rVXLV260K2srEzYWPqxsWuZMWNGtE5IOJ10v9b02l4GxaBBQ+Vr1sR6Hj+eYDtgwGAFAHzzzddNhEKh8cCBhCvZ2ZnC6dM/9nV0dKquO9bfv4Vq0qQp+TY2toZNmzY4RUfP8t679+erIpGIrbnOf9p99tmMjPnzv83YvPk7x5kzP2u+a1f89W++WZwRFtbP6pNPIjN79OhVAdQMraqLNyQkVLl8+WKv9PQ0M2/v5tU15zNR+vrrbxQCwJIlC9wLCvLNNm/elsznC9gZM6Y3W716ueu0aZF5Fy/+ZRMQEFheV2m4n3PnfrXbsGFLikhkbrx27Yr54sULvObOXZAWENBOtW/fbumMGZ8237HjwHWBQMBOn/6/5p07d1XExCy5nZ+fJ4iI+Ih4eTXV9OoVUr5q1XLXwsICsx079l9TqVTcadPCfevSGDw4rGzbtq2uCoWCZ2dnZ9Dr9Th79owkJmbxrfvljXp0ps5xeCwYhikGcBnA6NpNowEkMQxz9zAl/waf7QG8CuBa7Xe3BvsCAHgBYJ5oximKoqj/vLCw4cWOjk56V1c3nb9/i0pfX6Jq3bptlUgkYrt27a5IT79lAQAjR46RWVtbG83MzNgPP5yan52dZa5UKuvHI7Vv30nRsWNnFZfLhUgkYnk8Hnv7dpqovLycKxaLDW3aBKjvl4/ExGOSkJA+pX5+LTQWFhbGDz74ML/h/pCQvrKEhCNSAJDLy3hXr162GTRoaFndfj8//4pevULK+Xw+goN7yysqyvkTJ35UKBAI2EGDhpaVlpYIFQrFPcdPvfnm24U2NjZGP78WGnd3j6p27TqUe3k11dra2hqCgjooU1MZCwBISDgsCQnpK2vTJkBtZmbGfvzx9Ly0tFTL7OxM4alTJ2zd3DyqBg4cIhcIBOw770wotrW109WlER+/z2HUqDGFvr5+Gj6fj8mTwwuysjLNc3KyGn1Qv9uQIcOK3d09dGKxxDB27PiCM2dOSgBAIpEaBgwYrLCwsDBaW1sbx4+fUJCcfP2OrrrevUNlzZv7VJubm7M9egSX3b6dZgEAx44liNu166Ds0qVrpZmZGRse/kk+h8N55PmVarWa+/vvZyW9e4eWCQQCtkuXrvKEhMNSoGYI259//m43ceJH+ZaWlkZ//5aa4ODed1Ryhw17vUwikRr4fD7ef39SkV6v46Sl3RLV7ffyaqauO7/vvvtBkU6n45gyz8DCwsLYvn0nxZEjhyQAkJ6eZlZQkCfq3TtUybIsjh1LsI+I+CxHLJYYrK2tjePGvVvw66+nJQBQXq7ki8XS+ut47dpV8z59egT07t0tcMSIwa0apjN27PhCsVhiMDc3Z/fu3eUQGjqgJCiog4rH42HEiFEyPl/AXrr0l2VS0kXL8vJy/pQpHxcIhULWy6upNjR0QMnx4wkSADh37oxk7NjxBWKxxODu7qELCxte31vn7Oyi8/dvUXnkyCExAJw6lWhrbW2tb9s28L6/Y9SjM3nQWe1bo8cC8AUguns/wzAjTYxqEoDNhJAo1HRrjauN/zCAKIZhLgD4oHYOgw4AB0AswzBHa4+fTwgJQs0ysVoAbzXshaAoiqKefyFB7rKH6Q14GqRS+/oHIqHQzCgWS+pnvpqZiYxVVVU8g8GApUsXuZ07d0ZcUVEu4HC4dcNO+La2tgYAcHR01DWMd86c+embNm1w2bRpg7unp1fVpElTchsb1lKnrEwmIMS/fr+bm/sdw3cGDw6TjRs3qpVKpeIePnxQQoh/pbOzS32adnbiO/JtbW2trxvWZG5ubgQAlaqSa2dn1+gSO/b2DvXHC4VmRolEesd5kctlgtp8Cn19iaJun5WVldHS0spQWFgoKC0tFtjb29fnm8PhQCqV1n8vKSkWrlu32mPDhrXuddtYluUUFhYKTBmu5OTkUh/Gzc2tWqGQC4Gah/WFC+d5JCVdtFGpKvkAoNFouA3nmjS8ziKRyKjRaLgAIJOVCBwcHOrjtbS0NFpZWT949vM9JCQctuPxeGxwcB8lAISGDpBFRn7iW1pawjcYDDAYDBxX17+vrbOzS/X169fqu/Y2bFjjlJBw2F6hkAsADjSaKp5cXlb/3Nbw/HK5XEgkUl1JSZHAlLz16dOvbM2ale5TpnxccPjwQUmHDp0UFhYWxpKSYr5WW8394IPx/n+HZmE0GjkAYGNjoy8rk9Wn0bp1m6pjx85cPnPmlPWSJTFeDdNwcXGtz19xcZHw9OkT0p9/jnes22Yw6DnFxcVCLpfLKhRyYZ8+PQLqU2SNHEL8KwBALpcLGsbV8HPteS2Nj9/nOHr02NKEhCPS4OA+z9X/K/81JlUcah/UzwDIRk3F4SoAW9S09ucCSDM1QYZhbgLo1Mj2AQ0+R9zn+LdNTYuiKIqiHqcDB/ZIzp//zW7ZstWpnp5NtEqlkjdoUO8A9o4B/He2UgcGBqkDA4PSdTod54cfNjpER8/0jo8/evVeaUgkEl1JSVF9y3teXu4drfCurm46Hx9SmZDws93x40elQ4aEFT+2Aj4EiUSqLSwsqB92pVKpuCpVJc/Z2VknlTroSkvP1eebZVnIZLL671KpvXb06LcKhg0bUXZ3vKYoKiqojys/P09oZyfWAsDGjeuc8vJyROvWbUpxcnLWX7t21Xzy5HdbsHdPsGi0PPa67Ows87rvarWaW1lZ0ehzEofDeWB8R48ekWo0Gm5YWP/apUNZGAwGzsGDByTjxr1TzOPx2Pz8XKGPD9HUlKmw/lz+8cdvVnv27HRevHhFqp9fiyoul4u+fXvccZ+VlpbWnwOj0YiyMpnAwcGptlJ0//z16NGrfNGiefxr166anzlzUjJ5cngOAEil9nqhUGjcvHnbDVdXN93dx7Vr175869bNriqVivug4UoNz5GDg6OuZcvWBR9+OPUfDb0XLvxp6eDgWL1nz6FG59La2dnpCgryhXXziAoKCu74fejbt78iNnZpk5SUG6JLl/6y/d//Psm9b+Gpf8XUoUqLAOxCzVwCDoD3GIZpBqAbalZXWvhkskdRFEVRzw+VSs0TCASsWCzRq9Vq7ooVi93uF16r1XL27dstUSqVPIFAwFpaWhrreinuJTi4T1li4jFpaupNkVqt5m7YsMbl7jB9+/aX7dy5zTknJ8s8NHSgorF4nrQ+ffqVnThxTHrt2lXz6upqzvLl37p5e/uoPD29tK++GqLMy8sxP3LkkJ1er8emTd85KpWK+pbqIUNeK9m+favLzZvJIgBQKpW8Q4cOiE1N++DB/Q55ebkCubyMt3XrZpfu3XvKAUCtVvOEQjOjra2dQS4v43333RpX08sTKr906S/b8+d/t9JqtZyVK5e6sizb6BO4VOqg53A4yMzMaHS+Sn5+nuD69as20dHfpG3c+OONmp+fboSFjShMTEyQ8ng8dOzYWbFuXZyrWq3mMkyK6MSJ49K641UqFY/H47JSqb1Or9dzYmOXuVRVVd0xvCwz87ZF3fnduHG9k0DAZ+tWNbK1tdPl5eXccy5NzdCpbvJVq5a5V1ZW8nv0eLUcqOm56N07tHTJkgUeJSXF/LqynDqVaAMAw4a9LrOzE+s+/XSqd0rKDZHBYIBGo+GkpFy/7xCpsLDhJb/8csjx4sW/LFmWhUql4h4/ftS2oqKCGxgYpBKJzA1r165yrqqq4hgMBqSk3BAlJV20AIBu3XqU/fjjFhe5XM7Ly8sVxMfvdWwYt0gkYjt37iqPjp7VzNvbR+Xu7mnSBHvq0ZhacQhAzTsX6mqXIgBgGOY3AHMAxDz+rFEURVHU82XYsBEyBwfH6tdeG9B2zJjhLVu2bP3A5TWPHftF+vrrg1v37t0t8NCheIcvv/wq437hg4P7lA8eHFb88ccfkZEjh7Rq0aKlCgCEQmF9C29o6ABFaWmJWceOXeQWFhYPnKh6L3PmzPCcM2eG56Mc26NHr4qxY8fnR0VFeg8Z0rdtQUG+2ddfx9wGalquZ8yYnb5hwxr3/v1fDcjLyxH5+vpV1h3bv/9AxahRYwpmz57RrHfvboFvvfV6yz/++M22sXTi4lY6f/TRBJ+G23r1CimLiPjId+TIsNbOzi7VkyaFFwDA2LHji7Taau7AgSEB778/zr9jx87KxuJsjJ9fC83kyVOz582b3XTw4L5tra2t9RKJpNGHUAsLC+OIEaMKwsM/8OvTp0fAxYt/Wv7xx29WISFdAwHg4MF90iZNvNQ9ewaXOzk56+t+3nxzXHFOTrZ5SsoN0eefz8zWaKq4Q4b0bTtv3uymISF961cL6tUrWBkQEFT+5psjWg8bNqC1UCg0NhzqBQBBQR0VJ04ck/Tr1yvwxIlj0ujomHSBQMACwJgxbxVu27bVpU+fHgEbNqxxaqwMoaEDyq5fv2rzyivd5A0neH/yyee5rq5u1R98MN6/d+9ugf/732TfzMwMEVDzkL569YZUD48mmsjIaT59+vQIHDUqrBXD3LSMipqb3lg6ABAQ0E4dEfFZ5rJl33r27dszYOTIoa2OHDkoBQAej4dvv12elp5+y3z48EFt+vd/NSAmZq5XRUUFDwA+/PB/BQ4OjtUjRw5tHRHxkW9jQ5EGDhwiy8nJNu/TJ5QOU3rCOKZ03xFCZABGMAxzkhBSCCCCYZhttfv6ANjPMMyL8OIPLwAZdFWl+3sZyvkylBGg5fwveU5WVWoKIPNR40lKupzp7u5d+uCQVEOpqTdF7733VsuTJ3+/2PABb9iwAa0iIj7Lqls5h3p5NFwa9lnn5XmQm5stHDfujZb79/9yxcbG5pEr0lSN3Nx0+8DAAK/G9pna45CMmhexAcDvACIIIT6EkCYAPgNwz1rmf0VVpRJ/JcbDlIoWRVEURf0bR478bFddXc2Ry+W82Nhl7kFBHRQNKw2HDx+043A46NatB600UC81o9GILVs2Ob3ySnc5rTQ8eaauqrQOQN060l8COAqgbq1lFYARjzlfz53ff96Ekrx02Do3h9jR41lnh6IoinqBxcWtdN69e/s/5i74+bWoXLVq/a2DB/c5LFkS48XlcuHv37IiMnJWdl2YCRPGkdzcXNHnn8/I4HKf6qrqFPVcUalU3CFD+ra1t3fQLl68MvVZ5+dlYNJQpbsRQqwAdAFgDuCP2vczvAi88IhDlY5uXQh5cS76vPkpJE6PNBz0hUGHffx30HL+d9ChShRFUdTTcL+hSqYuxypiGEZT951hmEoAxx5P9iiKoiiKoiiKet6ZOlRJSQi5CODX2p+zDMM8k+XfKIqiKIqiKIp6+kytOIwB0B1AbwDTAIAQkow7KxIvxws36NxoiqIoiqIo6iVkUsWBYZg9APYAACHEGkBXAD0AhACYhJrHaVMrIS8mE94SSVEURVEURVH/VQ/1sE8IsQDQEUDn2p9WACoA/Pb4s0ZRFEVRFEVR1PPCpHXcCCHfEkLOA1AA+BFASwD7UdPzIGYYZsCTyyJFURRFUS+6CRPGke3bf7Q3JWy3bu2Dbt9ON2tsX1hYv9Znzpyyfry5a1zDtOLiVjpHRUU2edAx1NNxv3vkcTH1XsvOzhR269Y+SK/X3zdcbm62MCSka6DBYHhseXzaTO1x+ARAFYA1ADYwDHP1yWXp+YPD5ecAACAASURBVMbSSQ4URVEU9dKZPDm8sO5zdnamcMyYEa1Pnfrjjrd53+3SpQsW33231jU1lbHicjlwcnKuHjw4rPj110fLHjUfv/32q/U333zd9ODBo0/kWWzFiiWue/bscObz+SyXy2Pd3d01H330cU779h1VTyK9x2HGjOlep0+flM6aFZ0eGjqgfvGeb76J9vj553jH//3v08zXX3/jkc/5w5owYRwJCQmVvfHGm3csP+3u7qlNTDyXZEoc97rOd8edlnbLLC5uhduNG9esDQYDx97eQdu7d6hs/Pj3i/LycoRjxoxobWZmZgQAMzMzY+fOXeVffBGVIxAIHumB1tQ3x/QDsBRAWwDnCSEyQkg8IeRTQkhHQgjvURJ/sdA5DhRFUdTDeZFbFql/58KFPy2nT/8fadMmoHL79r3Xf/nl1OVp0yKzzp//3fZZ5+1BunTpJk9MPJd05MiJy23aBFTMmTPT+1nl5UGt+HWcnV2qExIOSxsed+7cr2JHR6fqJ5a5Zywj47bZhx++5+/g4KjdtGlb8rFjv16Ojo5JT029aVFRUVH/bJ6QcDopMfFc0ubN228wTIrV1q2bHB41TVMnRx9FzduiQQgRomaeQw8AQwEsQM3bo20eNRMURVEU9SyFhfVrPXDg0OITJ45Ji4qKzLp27V4WHh6RN2fOzKYMk2Ll7e2jiolZki4Wiw3TpoU3S0lJttbptBxPT6+q6dO/yPLza6EBalo+hUIzY3FxkTAlJdl69ux5aTqdjrt2bax7WZlMKBKZG8LChhe9//6kovvlZ9261U4HDux14nCAsWPfyY+NXdpky5Yd15VKBW/GjOk+Bw8eu8Lj1TwXHDlyyG7Llo2u27btSV6xYolrVlaGSCAQshcunLezt3eonjt3Yfrx4wniAwf2OgkEfHbatMjMnj2DyxtLt1u39kETJ07J3r9/t5NSqRAMGhRWNHToa7Lo6JlNc3KyzQMC2innz/82QygUsgCwbdtW+127tjmrVJV8Qvwrv/giKsvFxVUHAKdPn7SJjV3qoVDIBT17Bpfd/b7ZnTt/ku7atd1ZqVQIvL19VF9+GZXl4dFEa8r1Sk6+bhkbu9RToZALOnbsrJg5MzpLJBKxcrmcFxUV2TQtLdXSYDBwCPGvjIycleXm5q4DalprW7ZsU3HlSpJNTk6WuY8PqZw7d0GGVGqvB4C9e3dKNm/+3k2j0fCGDn2tsGGaK1Yscc3PzzOLiVmcMWXKRAIAoaE9AwFgwYKlqXe3yK9evcK9V68Q2cSJH9XH07ZtoLpt28DbALBr13bpkSOH7L//fivT8Pxv2bLjerNm3tWJicds775vRo9+q2TGjM989Ho9JySkayAAbN2667pEItUvXrzA/bffzogB4JVXesinTfs818zMjK1ruR48OKxo377dzlwulw0Pj8gWCITs6tXLPSorK/hhYSMKG/ao1OHz+RgwYLBsz56dzqWlJXx7ewe9UqnkffvtN+6XLl2w5XA4CAnpWzp16if5PB4PQ4f2ax0d/U1627aB6n37dksWL45pumHDlht+fi00O3b8aP/77+dsly1bnX7p0gWLFSsWe+bl5YmEQqGxS5euis8+m5FTd1/V3IcfZe/fv8fJYDBw9u8/cm39+jin/fv3OHE4wLhx7+bdndcOHTorTp9OlMrlcp5YLDacOpVo26SJV1VVVVV9I7nRaERc3EqXo0eP2Gu1Wm5AQDtlZOSsHFtbW8ODrr/RaMT69audf/nlsL1area3bt2mfMaM2VliscTk1oG7e6oedD8+yNq1q1x9ff0qIyNn1a9s6uPjW71gwdIMACgvV9zRsO/g4KgPCGhXnpmZYW5qnu/2UO+qJ4RIAfQHEFb70xk1TfEvx1KsFEVR1GOhvXFcqto3hzzJH+2N49IH5+RvZ8+eES9fHpe6devO6xcu/GkXETHFZ9KkKbmHDh2/zLIstm7d5AgAnTp1Ue7Ysf/aoUPHr3h7+6ijo2c1uzOe05K3336v4PjxX5M6duxcuWTJgiaffPJ51vHjZ5N++GHnjY4dO9/3FeAnTx632b9/j/PixStSd+6Mv3758qX6MdaBgUFqKytr/enTJ+ob6xISDktDQvrUD8O4ePGCXb9+A2S//HIqqWlTb/Wnn071NRqNiI9PuDp69Lj8pUsX3Xec/oUL5202bvwpOTZ2Xcr+/budY2Kim3z11bzbe/b8fDU7O8s8Pn6fpPZ8WW/evMFt9ux5t+Pjj15xdHSqnjnz82YAIJOV8r/+epb3+PHv5x8+fOKKq6ubJjX1plWDPNtt3/6jy7x5C9MPHz5xuVWrNpVRUV80u1eeGjlH0iVLYlN37Nh/LS8vTxQXt9IFAFjWiP79B5Xu2fPztT17fr4qFAqNixbN92x47OnTJyQzZszOOHjw2GW9XsfduHGDEwAwTIpo1arlTSIjZ2XExydcKS9X8svKyoSNpR8bu5apKUdNS+7dlQa1Ws29dYuxCg7uLTe1THdr7L6xtLQ0zpu38JadnViXmHguKTHxXJKLi6tuzZqVLgyTYrlx40/JmzZtS75166Zl3TkBAKVSIdBqtdz4+ISrY8eOz1++/NsmR48elmzatC15+fK4mzt3/uSalZX5j7JqtVrOwYMHpFZWVno7O7EeAL766gsvHo+HnTsPXN+48cfkpKQLtjt21MxdadGiVcVff523BoDLly9ZOzg4Vl+48Gft9ySrNm0CKgCAx+MhPDwi58iRE5fj4jbcvHIlyfrHHzff0Qp+7tyvdhs2bEnZtm3v9ZMnj9vs27fL+dtvl6fu2HHg+sWLf/2jsVooFBo7dOis+PnnAxIA+OWXn6V9+/a/Y3jS7t3bpYmJR6XLl8cxu3bFX9NoqngxMdGewIOv/+bN3zv+9ts5u9jYdUx8fMIVKytrwzffRHviX7rX/WiKq1cv2/Ts+arJ91hBQb7g0qWLNi1btqp8tNyaPjl6DSHkBoBiALsB9ARwFsDrABwZhmnxqBl44dApDhRFUf9JYWHDix0dnfSurm46f/8Wlb6+RNW6ddsqkUjEdu3aXZGefssCAEaOHCOztrY2mpmZsR9+ODU/OzvLXKlU1rfstW/fSdGxY2cVl8uFSCRieTwee/t2mqi8vJwrFosNbdoEqO+Xj8TEY5KQkD6lfn4tNBYWFsYPPvgwv+H+kJC+soSEI1IAkMvLeFevXrYZNGhoWd1+Pz//il69Qsr5fD6Cg3vLKyrK+RMnflQoEAjYQYOGlpWWlggVCsU9hxi/+ebbhTY2NkY/vxYad3ePqnbtOpR7eTXV2traGoKCOihTUxkLAEhIOCwJCekra9MmQG1mZsZ+/PH0vLS0VMvs7EzhqVMnbN3cPKoGDhwiFwgE7DvvTCi2tbXT1aURH7/PYdSoMYW+vn4aPp+PyZPDC7KyMs1zcrIafVC/25Ahw4rd3T10YrHEMHbs+IIzZ05KAEAikRoGDBissLCwMFpbWxvHj59QkJx8/Y7Jrb17h8qaN/epNjc3Z3v0CC67fTvNAgCOHUsQt2vXQdmlS9dKMzMzNjz8k3wOh/NIf/UVCjmPZVk4OjrpHhy6cQ9z35w6dUIybty7BQ4Ojnp7ewf922+/l3/q1N8VZx6Px06eHF5Qdw9UVFTwR44cU2xtbW3082uhcXNzr0pJuWFRF/7338+K+/TpERAS0rXdsWO/OERFfX2bz+ejuLiIf/nyJdvPP5+ZbWlpaXRwcNQPHz6q6OTJRAkABAQEVl69etkaAJKTr1uNGvVm4ZUr9d+t27fvWAnU9LwEBXVU8fl8eHp6aQcMGFxy5UrSHddp7NjxhWKxxGBubs4mJh6TBAf3LfX3b6mxtLQ0Tphw5+9EnQEDBsuOH0+QKhQKXnLydeu+ffvf8VCdmHhUOmzY60VeXk21VlZWxg8//F/ub7+dFev1+gde/8OH4x3ef39Snpubu87MzIydNGlK/vnzv4tNHUp1L/e6H02hUlXy7O0dHniP9e8fHNCnT4+A118f0kYkMjP26zfokSu0pk6O9gOwFzUve/uNYZhHrqm8qOgMB4qiXkQsawSrVoKtUoIr8QCH+3xMSRO27C0Ttuz91CYrmkIqta//AywUmhnFYkn9E4GZmchYVVXFMxgMWLp0kdu5c2fEFRXlAg6HywJAWZmMXzfcwdHR8Y4/5HPmzE/ftGmDy6ZNG9w9Pb2qJk2aknu/iaZlZTIBIf71+93c3O8YvjN4cJhs3LhRrVQqFffw4YMSQvwrnZ1d6tOsaxmuy7e1tbW+bliTubm5EQBUqkqunZ1do0Ms7O0d6o8XCs2MEon0jvMil8sEtfkU+vqS+omoVlZWRktLK0NhYaGgtLRYYG9vX59vDocDqVRa/72kpFi4bt1qjw0b1rrXbWNZllNYWCgwZbiSk5NLfRg3N7dqhUIuBGpa+hcunOeRlHTRRqWq5AOARqPhGgwG1J2DhtdZJBIZNRoNFwBkshKBg4NDfbyWlpZGKyvrR3oqtLMTGzgcDoqLiwQ+PkTzKHE8zH2jUMiFbm5u9WP53dzctXK5XFD33crqn/eAg4PDHde1qkpd35jcpUs3eUzM4gyZrJT/+eefeCcn37B45ZXuFbm5OUKDwcAZMiS0bV1YljVypNKaa92hQ+eK775b615YWCAwGo2cAQMGl/3002bX7OxMoVqt5rVq1UYN1EzoXb78W4/09DRLrbaaazQa0bRpszsqRi4urvXXouZ3wq++7B4eno3eI506dalcsGAuf926VS5BQR0V5ubmd1T8ysrKBA3j9fDw1BoMBk5xcZHgQddfJisVzpkzo3nd7zwAcDhcFBcXCfAv3Ot+5PH4rMGg/8fjp15v4PD5fLYmj1aG0tKSB6Z/5MiJy3w+H1VVVZyVK5e4TZ060XfTpm03HyW/ps5x6PUokVMURVFPF8uyMJZkgK0qhz73Ogy512FU1gzVNR/wKfjurZ5xDl9sBw7skZw//5vdsmWrUz09m2iVSiVv0KDeAewdA/jvbKUODAxSBwYGpet0Os4PP2x0iI6e6R0ff+8VcSQSia6kpKi+5T0vL/eOVnhXVzedjw+pTEj42e748aPSIUPCih9bAR+CRCLVFhYW1C+HqVKpuCpVJc/Z2VknlTroSkvP1eebZVnIZLL671KpvXb06LcKhg0bUXZ3vKYoKiqojys/P09oZyfWAsDGjeuc8vJyROvWbUpxcnLWX7t21Xzy5HdbsHdPsGi0PPa67Oys+rHfarWaW1lZ0ehzEucBL4W1sLAw+viQypMnE8Vdu/ZodGiaubm5sbq6uv5hvaio8I607nXfNNYLYmcn1ubl5ZnVzbXJz88TisXiR+7tqCOV2us//3xm5uTJ77YYNGhomaurm47P57N1D6J3a9bMu1ooNDP+9NMWx5YtW1XY2NgYbW3Fut27dzj4+flXcrk1xV20aF6TZs181PPnf3vb2trauHHjesdffz0tbhhXw3MskUh0RUV//07k5ubcs2eqV6+Qsl27trksXLiUuXufRCLRFRTk3xEPj8djHR2ddA+6/hKJVPfZZzMyOnbs/I/KW3b2P4d5/Vtubm7aiooKfmVlJdfKysoI1PwelZaWCF1dayo/bdoElJ85c0ps6ipd5ubm7JAhw0rj4/c5yWSlfFPnUjRk8hwHQogZIWQyIeQ7QshRQohP7fZRhBD/h4jHlxDyOyEktfZfn0bCvEMIuUoIuUwIuUYImdpgH48QsooQkk4ISSOEvG9q2hRFUf9FxooSaG8kQh0/H5Ub3od6fzSqEpZBd/MUYGYBYcfXYT5gOnhuLZ91Vl94KpWaJxAIWLFYoler1dwVKxa73S+8Vqvl7Nu3W6JUKnkCgYC1tLQ0NmyxbExwcJ+yxMRj0tTUmyK1Ws3dsGGNy91h+vbtL9u5c5tzTk6WeWjoQEVj8Txpffr0Kztx4pj02rWr5tXV1Zzly7918/b2UXl6emlffTVEmZeXY37kyCE7vV6PTZu+c1QqFfUto0OGvFayfftWl5s3k0UAoFQqeYcOHRDfO7U7HTy43yEvL1cgl5fxtm7d7NK9e085AKjVap5QaGa0tbUzyOVlvO++W+NqenlC5Zcu/WV7/vzvVlqtlrNy5VJXlmUbrSFIpQ56DoeDzMyMe75HYPLk8NyTJ49L16+Pcyork/EA4Pr1a+bTp3/cDAD8/Fqo8/JyzK9du2qu0Wg4a9asrM/r/e4be3sHfWVlBb/h8LgePV4t27p1o0tpaQlfJivlb978vUvPniGPpUfPx8e3uk2bQOXGjeudnZ1ddG3aBCgXLpznUV5ezjUajcjIuG32++9n6+evtGzZquLw4YOOAQHtKgCgdes2FYcPH3Ssm98AAFVVVTxLS0uDlZWV8dYtRvTzz/GO98tDcHCfshMnjtkzTIpIrVZz16+Pu+d1feut8UXz53+b2rlz13+Mjnn11T5l+/fvdsrKyhRWVlZyV69e4dalSzc5n89/4PUfMGBw8fr1q93rhtOVlpbwExIO290rHwaDnqPRaOp/dDrdQw1ecXf31DZv7qNaunShe2VlJbe6upqzfn2cE4/HY9u161AJABMnfpSfmnrTauHCee51Fc/bt9PNIiOnNW1sOGJ1dTXn0KF4qa2tnU4ikT5Sb5pJPQ6EEF8AxwDYArgIoBeAurFo3QEMBDDOxDTXAFjFMMxWQshYAGsBBN8VZg+ATQzDsIQQawDXCSGnat8f8SaA5gB8AEgBJBFCjjMMk2li+v8SneRAUdTTZ9CoYFQrYci7AaOiAEZlIYzyfLCaSrBVyppAQgvwfbqA59AUXFsncKWe4JrTBe8ep2HDRsguXDhv89prA9paWlrpx417N//o0SP3Xdrw2LFfpHFxKzyNRiPHxcVN8+WXX2XcL3xwcJ/y1FSm+OOPPyJcLod9442xBWfOnJQKhUJjXZjQ0AGKVauWNenYsYvcwsLCeL/47mfOnBmeAPDVV/OyH/bYHj16VeTkZOVHRUV6q9Uqvq+vX+XXX8fcBmpaqmfMmJ0eG7vMc8mSBV49ewaX+fr61T/I9e8/UFFVpeLOnj2jWWlpiZmFhYWhTZvA8kGDhv5j7HVc3Ern69evWq9atf5W3bZevULKIiI+8pXL5YIOHTopJk0KLwCAsWPHF0VFRTYbODAkwM5OrBs+fGThhQt/3vPhriE/vxaayZOnZs+bN7tpdXU1b+jQ1wolEkmjQ2IsLCyMI0aMKggP/8BPrzdwYmK+vaXT6TkzZkz3qVunv0OHTqpFi5Yz33231m3nzp9cuFwunJ1dNEOGvFYCAM2b+1SPHDkmf/r0qb5CoZB9550JuceOJdTfS/e6b3x8iKZr1x5lI0cObW00GrFly/YbkyeHFyxeHMN7++3RLQDglVe6ySdPrjknj8Obb44rmj79Y9/3359UEB0dk7ls2SK3N98c0UqjqeI6OjppR416sz6ttm0DK86d+1XSoUOnSgBo1659RXz8Pqd27drXX//Jk6fmLFmyoMm+fbucvbyaqrt371lWNxeiMcHBfcpv3WKKIiKmEC6Xw44b927e2bOnJY2FFYslhu7dezbayzNy5OjS0tISQXj4B346nY7Ttm278s8/n5kNPPj6jx//fjHLspyIiCm+CoVcYG1to+vevae84XsjGoqLW+kZF7eyfvJ0t249yz78MPwfq0Hdz9dfL0hfunShR821NqBpU2/1ggVLbolEIhYAmjZtVh0buz5lzZqVbuPGvdHKaDTA3t5R27dvv1Jra2tD3apKdat/cbk8tkkTr6q5cxekPajX7F44pnTfEUJ+AWAJYDCASgBaAO0ZhrlECHkdwAKGYR64GgIhxBFAKgApwzCG2vc/yAD4MAxTco9jnAAkAQhlGOYaIeRnABsZhtlduz8WQBbDMItMKK8XgAyZrBJG48NVAI79tBhlhVnoPfoTSF28HurYF42DgzVKSu676McL72UoI0DL+aJiWRZGeS6Mshzo85JhKLwFtqIEYP9+PuRY2IFr6wyOuQ14jt7gOTcH174pONyHWizPJFwuB1KpFQA0BZD5qPEkJV3OdHf3Ln1wSKqh1NSbovfee6vlyZO/3/GysWHDBrSKiPgsq0ePXv+dm5+iqGcuNzfdPjAwwKuxfaZOju4O4HWGYRSNvOytCMA/ulHvwQNAHsMwBgCorTzk126/o+JACBkC4BsA3gC+YBjmWu0uTwBZDYJm1x5vsto/gA9FwK8ptp2dBRwcnsqb7p8pWsb/DlrO559BXQ5V6p9Q37oAQ0UZdGUFMFbXzBPkiixh7uEPnndb8G2kMHPxhqhJS3D5j31ILfUcOXLkZ7vg4N5KtVrNjY1d5h4U1EHRsNJw+PBBOw6Hg27dGh8/T1EU9SSYWnHQALjXyyLcADz28ZUMw8QDiCeEeALYTwg5zDDMPya6PIpH6XHQ6WsWn1Ao1OCZ/7f/n/6vtd425mUoI0DL+TxijUaw1ZUwKgqgz0qCIecajPJ8AGxtL4ITuJ6B4Fvbg+/VDlyJOzhcXn0Z9QBU8moAT+9lqA16HKjHJC5upfPu3dv/0ejm59eictWq9bcOHtznsGRJjBeXy4W/f8uKyMhZ9UOJJkwYR3Jzc0Wffz4jg/sEepgoiqLuxdSKwzEAXxJCjqNmqBIAsIQQMwDhAA6bGE8OADdCCK/BUCXX2u2NYhgmmxDyJ4BBABjU9DA0AfBXbZC7eyCeKFOGdlEU9XJjq1UwFN8GwAJcPoyyLOjzUmBUFoHVVADa2lUHuTzwXP0h9O4InrMveC4EHA59EHwZTJ4c3uibeuusXr3h1r32rV+/5bE0olEURT0sUysO0wGcA5CGmkoECyAKQEsAQgCvmRIJwzDFhJDLAEYD2Fr7b9Ld8xsIIf4Mw6TUfrYH8Cpq3iMBALsATCCE7EXN5Ogw1AyloiiKemZYvRZslRLaa0ehu5F4x3wEAODaudT0Jri1AMfCFlwbR/A924JjZvmMckxRFEVRD8fU9zjkEELaAvgEQAiAdNTMa9gFYAnDMA+z5NckAJsJIVEA5KhdjYkQchhAFMMwFwB8QAjpC0CHmnevxTIMc7T2+B8AdAJQ1xoTzTDMfVeooCiKehIMZbnQJZ+AUZ4HQ8HfjcD8Zh0h8HkFqB1GwrG2B8/O5FUhKYqiKOq5ZGqPAxiGkQOYVftzB0JIR4Zh/jQxnpuoefC/e/uABp8j7nO8AcBkU9J6Eh51+SqKol58rKYS+sxL0KX9DmNFac1KRxwuuLbOELTpB66VPbh2zs/VS9ZYlkWJogSyqjL4uRD6fxhFURT1yEyuODSGEDIINcOYugH4x4sm/ovoHAeKenkYVXLobp6GoSgNbEUpjBWlgFEPjpkVuE7NwWveGQL/XuBaSZ91VgEAWm01MgszkV2QBUW5HFWaKugqNODra3o+lJ2V6Oz7j3YbiqIoijLJPSsOtROXP0fNUCIPABkA5jAMs4sQ0g/AIgAtUDNhefyTzypFUdSTZdRUgFUWwSDLgT7jLxjyGYA1AAIRePZe4Ds0hYB0r5nEzH12bSVGoxFarRYKZRnSitJRWFyASrkSnCojOKjpUTCChZ5nACx4sHSQwMHGAe2aBj6zPFMURVEvvvv1OEwDMBdAAoD9qFnJaAshpBNq5jowAEYB2MMwDG2GpyjqhcIajWDLi8Hqq2t6FfJTYFQ0eNEqlw9+s/YQBgwET+p574ieMKPRCGWFElm5DPLyi5BXmIMqpeqOl9gbOEZo+XrwJWZwdfOAnbUdWnm2gpXAkg5Nop4bEyaMIyEhobI33njzgS8B7NatfdCWLTuuN2vm/Y91h8PC+rX+5JPIzKfx4ruGacXFrXQuKMgzi46OeWorOd4tNzdb+Pbbo1sePXomicfjPdQ5fd6tWLHENT8/zywmZjGdt/ocu1/FYTyAxQzDTK/bQAgZDeBHAAcBDGcYRv9ks0dRFPX4sAYdtFeOwFCUBkPO1Tv2cZ2aQxgwEFxpE3AlbuDauTyTpVENRgNuyzKQmZcBpUKBilwZeMa/86Hj6qEwV0NoLoKdrRiOto7o2Lw9bEW24NKlXCnqiWm4fG52dqZwzJgRrU+d+uOOt3k3NGPGdK+zZ89I+Hx+fTXf0dG5etu2PcmPmgd3d09tYuK5pEc9/n7S09PM1qyJdbt+/Yq1Tqfj2tra6QIDg5TvvDOh0M3NXfck0qRePPerOHgBOHTXtoO1/y6jlQaKop53hrI8sGo5WJUc2uQTMJbUNmRxuOBY24Pv0RZcW0fwm7QD18bhqebNyBqRXHQTubJclJaWQCUrB7eKBY/lgG+sGQbFgkW1uQF2zmKIJTYQWFnB3kKCllI/8J7hUCnKdAaDATwevVYvq7CwEYUREdPzn3U+HuT27XSzDz98z//VV3uXbtjwQ7Kbm7uupKSYf+jQAcnFi39Zubm5y+8+Rq/X416VJuq/635XXASg6q5tdd+VTyY7zy/a3U9Rzz9DSSb0GX/BqFbCkJ8CtvLOlaI5lmII/HtB2DoUHIHoqeWrvLoClzMvo7xSCYVCAW1FFXQaLcy1gvowZlwuWAsuIODCTiyFs9QZzk5ucBe7gcvhvlBvx34RhYX1az1w4NDiEyeOSYuKisy6du1eFh4ekTdnzsymDJNi5e3to4qJWZIuFosN06aFN0tJSbbW6bQcT0+vqunTv8jy82uhAWpamYVCM2NxcZEwJSXZevbseWk6nY67dm2se1mZTCgSmRvCwoYXvf/+pKL75WfdutVOBw7sdeJwgLFj38mPjV3aZMuWHdeVSgVvxozpPgcPHrtSVyE5cuSQ3ZYtG123bduTvGLFEtesrAyRQCBkL1w4b2dv71A9d+7C9OPHE8QHDux1Egj47LRpkZk9ewaXN5Zut27tgyZOnJK9f/9uJ6VSIRg0KKxo6NDXZNHRM5vm5GSbBwS0U86f/22GUChkAWDbtq32u3Ztc1apKvmE+Fd+IaavAQAAIABJREFU8UVUlouLqw4ATp8+aRMbu9RDoZALevYMLrt7bZGdO3+S7tq13VmpVAi8vX1UX34ZleXh0URryvVKTr5uGRu71FOhkAs6duysmDkzOkskErFyuZwXFRXZNC0t1dJgMHAI8a+MjJyVVddiPmHCONKyZZuKK1eSbHJyssx9fEjl3LkLMqRSez0A7N27U7J58/duGo2GN3Toa3e8oK/hUJopUyYSAAgN7RkIAAsWLE1t376jypS8A3/3WHz00f+yfvxxsyvLAsOGjai/Ly5dumCxdOnCJgUF+SKhUGjs0eNVWWTkrNz79XQYjUbExa10OXr0iL1Wq+UGBLRTRkbOyrG1tTXUHffxx9Mzt2z53lWrreYOGfJa0YcfTi0EgHXrVrvWnqvcuvgcHBz177wzobju+2+//Wr9zTdfNx00aGjxgQN7ndq2DSifN29Rxvr1q51/+eWwvVqt5rdu3aZ8xozZWWKxxAAAFy/+aRkbu8wjNzdXZG/voA0P/zjnlVe6VwBAVlam8Ouvo5pmZt62aN7ct9Ld3aN+WNqUKR8079ChU/nbb79Xn/7o0a+1GDfuvfz+/QcqTD3P1OP3oKricEJI+wbfuagZWfs6IaRzg+0swzBxjz13FEVR96DPvwlD7jWwmgoUstWoyk4BW9XgWYjLh6B1KHjOPuBa2IHr6P3EGgBYloVSWw6NXoMKTSX+TDkPZZkcHD4XHCMglAN89u9WZw7XCCGfD8fm7nCWOMNF4gpHB6eXqoHidO456fmCi/ZPMo1OLkGlPd27mvyeobNnz4iXL49L1ev1nPfee6tFRMQUi8jImZk+PkQzdeokn61bNzmGh0cUdOrURTl79vxMoVDILl68wD06elazn376e/jJ2bOnJfPmLbrVvn3HNK1Wyxk+fFDrWbOib3fu/EqlXC7n5eRkmd0vHydPHrfZv3+P85IlKxlPTy9tdPSsJnX7AgOD1FZW1vrTp0/YBAf3KQeAhITD0pCQPvXlvHjxgt3s2XPT5s5dkDFrVqTXp59O9e3bt39JfHzC1d27d0iXLl3UpGfP4Gv3Sv/ChfM2Gzf+lJyfnyucNOndFikp162++mrebbFYYpgwYZxffPw+yYgRo2Rnz56x3rx5g9vChctuEeJftWjRfPeZMz9v9t13PzAyWSn/669neUdEfJbZt29/xQ8/bHQ4evSIQ+/eobLaPP+fvfMMjKJq2/A920vKlvTeJwlFCF2qKfQughV5VQSkfAZEkQ4iVXqoooCggII0BUPvWMBoRGBCSUJ62ZZsy+7OzPcjZEkghCBFhLn+wMzpM5vd85ynKbZu/dp37tyFV8PCIqxr1qzwnTr147D167+5XJ93dfToIfWiRSkZcrmcGTt2dOSqVct9k5PH57Msg27depZ26pRwnaZpTJ8+MWTBgtlBS5asvFbV9vjxI6oFC5Ze8fcPsI0ZMyxq/fp13h98MCGPoi5JVqxYGjxr1vwrcXHNTYsXz/fXarWi2sZPSVlDvfrqgEapqcfTHuTUPS3tvOu2bbsvZGdniseNGx0VFRVt7tChU/myZQuD+vUbUNS//0Ct0WjkUdRF6b362r59q/rw4QPqpUtXUR4eno4pUz4KnTt3ZtCcObd8BtLT/3DZtm3XhWvXrkpGjRoak5jYWR8VFW1NT//D7X//G5pbV/8AYDDoheXlZfydO/elMwyDjRu/9Dpz5rQiJWUt5eHh6ZgzZ2bQnDkzg+bPX5KZn58nnDjxw8gPP5yU+cILCYaTJ4+7zZgxJfzrr7+74OHh6Zg27eOw6OhY44oVn2f8/vs5+ZQpH0U2b95KDwBdunTXbN++1btKcLhw4S+pVqsVJSQkPXMH108a9/q0j7/L/Y9uu2YBcIIDBwfHI6Ei7Qc4rpwBU1YMMHdaSVYZ3wqi2kHcvD8IueKR+ieU24ww2k0w2oxIK0hHdl4W2DI7eCCgsMghYgSQE4LKb0YC4LuIEBwcBl9PP/h6+EIudXlkc+P45/Tt+2Kxl5e3AwBiYmKN7u4Ke6NGz1kAoG3b9vq0tPOuADBw4KvOTfp7743J79kzsYnBYOC7u7vTANC8eSt9y5atTQAgkUhYPp/PXr9+VRIb29CsVCpppVJprmsehw8fVCUkJJVWaTHeffe9/FOnjquqyhMSOmtSU/er4+OTynQ6LT89/Q+3CROm3Kgqj46OKe/UKaEMAOLjE3W//npWOWzYyEI+n4+ePftoV6xYEqzX6/kKhYKubfzXXnuz0M3NjXFzi7UGBARa4uJalIWEhNoAoFmzFoaMDEoGQJOauk+VkNBZ07hxEzMAvP/++LwePRKa3LiRJfrtt19d/f0DLT169NYBwP/+N7R4587tPlVj7Nmz03PQoFcLo6KirQAwYsTogsTEbT45Odmi+mgdevfuVxwQEGgHgNdfH1KQkrI4MDl5fL5Kpaa7d+/lPJEeMmRowbhxo8jqbRMTu2giIiIrAKBDh3jt2bOnFABw8GCqMi6uhaFNm7ZGABg9emz+/v0/eN1rLnWxe/f3Pvv27XX20aJFK/3s2Quyqq6HDh1RIJfLmdjYhpb4+CTNwYM/qTp06FTO5wvY3NwciUZTKlCrPRzNmt1bm3H48AF1v34vFVW9q/fe+7/cd955o4HD4XAKDsOGjcyXSqVsw4aNLEFBwZbLly9Jo6KirUZjucDDw8Ppx/DVV196btq0wZ9haKJt2w7amTPnZAMAQRDsqFHJ+WKxmAWAffv2eI4ZM+5GlUZn+PBR+YMG9W3kcDgy9+7dpY6La2ao2ux37PhC2datm03Hjh1xb926Tfn169fkK1asyxCLxWybNm2NzZq1cL63pKSu+uXLFwVfu3ZVHB4eUfHjj3vUbdu211Zpujj+Pe4qOFAUxXnZcXBwPHZY2g5GmwfH9V9B6/IAhw10/iUAgCC8NVizHkx5CQiJC0TP9QDfJxJegX4o1d5uWfnwyDbkIC3nD5QxJhRqCsHoKiBk+BAwPLhb5fCGHABA8HkQSIRo2rQlyLBoMAwDgiDA4/17X6c2mkGeuQJGuwMxChcIeE+GVqNjQFvN/WgDHgdq9a2Nk0gkZpRKlVNKFYsljMVi4dM0jcWLF/ifPn1CWV5eJiQIHgsAWq1GUCU4eHl51XAknTFj9rUNG9b5btiwLiAoKMQyfPio3LrMWrRajZAkY5zl/v4BNTbSvXr11QwePKihyWTi7du3V0WSMUYfH1/nmAqFssa8XV1dHVVmTVKplAEAk8nIu5vg4OHh6WwvEokZlUpd47nodBrhzXmKoqJI52bPxcWFkctd6MLCQmFpabHQw8PDOW+CIKBWq53XJSXForVrVwauW7cmoOoey7JEYWGhsD6Cg7e3r7OOv79/hV6vEwGA2WzmzZ//aWBa2nk3k8koAACr1cqr7mtS/T1LJBLGarXyAECjKRF6eno6+5XL5YyLi+sD+XP26dO/Th8HP79b79bHx7ciKytTAQATJ07LWrMmxe+11wY08PT0tr355lv5iYld6jxt12q1Ql9fP2d/gYFBNpqmieLiIqdNpLe3j3PtYrGYMZvNfABwcXFxlJaWOOsNHvxWyeDBb5UsXrzAr7S02Kl1cXV1c0gkEufmXaMpFc2YMSmi6u8AAAiCh+LiImFRUaHo7NkzyqSkDu5VZTRNE02aNC0vKioUyuVyh1wuZ6rKvLy8bSUlJSKgUuBu27a99scfd6tHjUrOP3nymGr69E+dWiOOfw/Oq+W+4YRdDo6HDWMpA6O5AUZfgIozX99RLohsC8nzr4IQy2ttT/Af7ldZBW1DliYb6ZfToC0shcTKd+ZH8IEMgAw8AR88Pg+BkaEID46Aj4/fHQLC43aKpVkWJRYbck1W6G0OXDWYkWeuAH3TuPytKH9EuMse65yeNnbv3qH65ZcziiVLVmYEBQXbDAYDv2fPxCY1k4MSNX4omjZtZm7atNk1u91ObNq03nPmzMnhe/YcSMddUKlU9pKSIudmLS8vt4a5jJ+fvz0ykjSmpv6oOHTogLp3777Fd/by6FGp1LbCwgKn2ZXJZOKZTEa+j4+PXa32tJeWnnbOm2VZaDQa57Va7WF75ZU3Cvr1G6D9J2MXFRU4+8rPzxMpFEobAKxfv9Y7Ly9Hsnbthkve3j6Ov/5Kl44Y8VZsfZK3qlQe9hs3sp0mQWazmWc0ltf65fKwzArz83NFkZGkFQCKigpFKlWlcBUeHlExf/6STIZhsH//D8pZs6aHt2nT7o+656+yFxTkO59Lbm6OiM/ns15e3vb8/NxaTa6qaNToubKTJ48rX3rplTqF+dvXrVKp7R9+OCmzSsNWHS8vb1uHDp00VdqK6uTkZItMJpPAZDLxqoSH4uJiUfX+e/Too5k9e0boc8/FGcViMXM/PiQcjw5OcODg4HissCxTmWSt+BoYXT5sf+67o44guCnE7YeAEMsfulBw53xY5BjzcKXgCrIKs2EyGcE30HC1Ve4fRHweRCoZYkJiwWMISKUyBAaGQCJ5fM7VtUEzLIqtNpgcNHKNVhSYK5BZboHRcesQ2UsqQnMPN5AKGZRiIbyldZrWc9QDk8nMFwqFrFKpcpjNZt6yZQv966pvs9mIH3/co4yPTzK4u7vTcrmcqX46Wxvx8UnahQvnhfTs2UcTEBBkW7dute/tdTp37qbZtu1rn9LSEnGXLv+Os2hSUlftnDkzwrp27amJiiKtS5d+5h8eHmkKCgqxyeUuhtWrlwft3/+DIimpq37Tpg1eBoPeeaLdu3f/kg0bPvePiYk1R0fHWg0GA//kyWNuPXv2uSN6T23s3bvLs1OnBL1MJmM2b97o2759Rx0AmM1mvkgkZtzdFbROp+V/8cVqv/qvp4tu5MihPr/8ctaladNmpuXLF/uxLFurhKBWezoIgkBWVqa4yuzpn7Bu3RrfadNmZd+4kSU6fPigx4QJkzMBYOfO71Tt23cq8/DwdLi6ujkAgMer+3PzwgtJ2u++2+LToUMng1rt4Vi5cpl/mzbtdPXxwRg69L38ESPejpk795OAwYPfKvLz87drNKWCnJxsqVQqrVUzBQDdu/cq/vzzlQG+vr6ZgYHBttLSEsH587+5dOnSXd+zZx/Nu+8OiT127LCufftOZXa7nfj9999cgoNDrYGBwbbQ0HBTSsoSv+Tk8Xlpaefl58//qqjycQCA5s1bmgiCwOrVywNeeCHxidJOPstwggMHB8djgXXYYPvjR9h+311rubBxN/C9QiEIeg6E4NFtcBmWwV/FF/HXjb9g0OpAm+2Qm0QQMnwIASggAgQElEGeCAoIRaOwRv+qqRHLsiizO1CuN6GozIxSqw2/l5Yh11Rzr+Iq5CPEVYoAuRiBLpX/Cv/FeT+t9Os3QHPu3C9u/ft3f04ud3EMHvxW/oED++uM5Xvw4E/qVauWBTEMQ/j6+lsnTpxWZ4Kr+PiksowMqvj990eSPB7Bvvzy6wUnThxVi0Qip1lHly7d9StWLAlu2bKNTiaTMXX1VxczZkwKAoBp0z69ca+6t9OhQ6fynJzs/KlTJ4SbzSZBVFS08ZNP5l4HALXawzFp0vRrKSlLghYtmhfSsWO8Nioq2ljVtlu3HnqLxcSbPn1SWGlpiVgmk9GNGzctq01wWLVquc+FC+muK1Z8fqXqXqdOCdrk5JFROp1O2KJFK/3w4aMLAOD114cUTZ06IaxHj4QmCoXS/uKLAwvPnftVUZ/1REfHWkeMGHPj00+nh1ZUVPD79OlfqFKpajWbkslkzIABgwpGj3432uGgiblzP7titzuISZPGR1bPs7Br13afH37Y5V11LRQKmZ9+OvZn1XWTJk3LBw7s3ZBhWKJfvwGFVdGufvnlZ/e1a1cF2mwVPA8PT9uECZOvS6XSOgWHgQNfKS0tLRGOHv1utN1uJ557Lq7so48m1+u9RkREVqxcue7SmjUp/m+99Vqsw+HgKRRKe5MmcYYhQ94pvFu7IUPeKWZZlkhOHhWl1+uErq5u9vbtO+q6dOmuDwgItH/yydyrq1YtD5gzZ2YYj8djw8MjTRMmVM5pxozZ1z/5ZEpot24vNImMJI2dOiVojEZjDTVtfHySZsuWTX49e/a5Wp91cDx6iPqo754iQgBkajRGMMz9rfvQlkXQFGQh4eX34eEX9ijm9sTwLIR9fBbWCPx762QZGnTe33Bk/wFacwOs2QDWZgYqKjXNPHUwBEGNQcjcwVMHQeAT9UDj3W2dNtqOCroCFzUUruozYbNYoc8qhsIkA5+9takWuIoR4B+EcP9wuLsrIZPJH7uwYLA5cK3MDIuDRmmFHboKO+wMi1yTFfbbvq98pCKQ7nJ4SEWQC/gIkIshF/AfWVQmHo+AWu0CAKEAsv5pP2lpf2QFBIT/5zPcPm4yMi5L3n77jQZHj56tEYKzX7/uDZOTP8x+HBmUOR4u9UkgxwHs2PGt+scf93h8+eVm6t+ey7NEbu41j6ZNm4TUVsZ9Wu+TZ0vO4uCoP4y1HIwmB6xRA+vJjTWiHxFSd/CV/uB7R0DYuCt4UrdHMofL2ivQWnXINNxAeakepfoS8BwEXG0SSBxCiGkh1HCBTO2KmLAG8PHyg5ubG4TCOs1/HwlmB42/dUbkmqy4VmaBtuKWP62Yz4OYR8BdJEQDpQu8pSIEql1hNVVALRHCSyJ6pkK3Povs3/+jIj4+0WA2m3kpKUsCmjVroa++wdy3b6+CIAi0a9eBExo4nkrMZjNvz56dnr179yv5t+fCcQtOcKgv3I80B8cdsAwN+4UDqPh52x1lfF8SwkZdIAyJe+jjGirKcK7oD/xWlAYbbUeAixcu5WYgyOAB9woZHAQNGctHENTONl4+PvD3DYSPjx88PLwe68a73F7pqGxy0Mg1WZFdboXRQYNmWYh4BALkEjRQuiBaIYdKLISb8E7twbOiJXtWWLVquc/27Vvv8F24Gdf+yt69Oz0XLZobwuPxEBPToLx6uNWhQweTubm5ko8+mpT5b5rRcXA8Ko4dO+w2c+bU8EaNGpf16dOf8294gqi3qRJJkhIAHQAEoDKrdHX+KwngQvBPTZW2LoYmPxPxg96Hpz9nqvRf51lYI/Bo1smyLBjNDdh+3wNH7gXAUWlrz/MIASFzh6hBAnjuPuC5PVD4cwCA1WGF2WHB2YJz2Jd5EC4COQiCgMVqhm+5En5GZa3t3JQKeKq9ERgQDA8PTwgEQohEj0erQDMs/tYbcUFrRAXNwMGyyCq3OOOxSfg8BLtIoBQL0dzDDb4ycb2EmH/zM8uZKnFwcHA8OzywqRJJku0A7ABwNwcwLgEcB8dTDGMth/3vI7BdOOD0UwAAiKQQNupSKSw8BEEBAPKMBdiesQelVi20Vh3AAg1KAtDCHg5e7QFOEBgYCJnMFSEh4fD2vuMQ95HAsiyMDhoXtEb8pTNCY7XBSjMgANgYFq5CPiR8HsR8Hlp5uSNaIYdaLIRSLASP02BycHBwcPwHqa+p0jIA1wF0BnCRoij7Peo/xXBODhxPPyzLAnYraM0NWFKXArZqiW7FchAgIEl8DwL/2Acey+Kw4mD2MRRbSvF36SXQDhpuFVIEWjwRaVbXqBscHAa9XgcvL2/ExDSCQlGpcXhcp/FlNgcogwlpmnLkGq1w3NTYSvk8BLtK4S4SgGVZRCvkiHKXcwICBwcHB8dTRX0FBxJAf4qi/rxnzXt1RJJRADYCUAPQABhMUdSV2+pMAfAyABqAHcBEiqJSb5ZtAJAIoErd/R1FUZ8+6LzuRVXyJw6OpxW2wgTL4VWgcy/cUcb3iYIgqi2E4a1ACB8sfwHLsrA4rNh+ZQ9+LfwdLFiABdwqpPCxukNtdoGIEUAoFMIOOyQSKUJCwtGiRZvH6pdQZnMg31yBfHMF8kxWXCszw3bTxFEtFqKx2gUKkRCxCjl8ZGJOSODg4ODgeOqpr+CQDsDnIY25GsAKiqI2kyT5OoA1AOJvq/MrgIUURZlJknwOwHGSJH0pirLcLJ9LUVTKQ5oPB8czCV2SBfvVs2BNOjCGAjCanBrlgtDmYAyFkHYbB568dl+C+yWnPB+7r+3DJW0GZDYxgllvuJukcGGlcFgqw6Xz+Xw0btoM0dENIBQK79Hjw8PioJFjsiKz3IILWiM01aIcCQgCjdUu8JKIEegiQZCLBHxOUODg4ODgeMaor+AwAsAGkiSzKIo6/k8HI0nSC0AcgKSbt7YASCFJ0pOiKGe4rSrtwk3SARCo1FDk/tOxHxzORInj6cH68zbY0/ffcV8U1wfCBgkPLVwqy7LIMxbgTMFvOF/0B6xWC4L1nmhliahRT+4uQ8xzLQEQCAoKfSxZmW00g0JLBewMizNFelzS3/LdCHKRoKHSBZ5SEcLdpHATCrjwpxwcHBwczzz1FRwOApABOEKSpA3AHcbEFEXVxzMyEEAeRVH0zTY0SZL5N+/fLU7vYADXKIqqLjSMJUlyGIBrAD6mKOpSPdcBAFXRQe4PU2U0MBe2HJ6ervff/j8Gt8anB09PV1jzr6Lou7mgjTWTsvq/sxAiD3+wDjt4YtlDG9NoM2HlzxtxPv8vKKxyuFRIEFahhtR+K7KRWCxGt27dIBAIEBwc/MBj3ut9OhgGmXozzhfqkFZogNZ6KyGsTMBHUqgXolQuiFK5QCZ8MiNVPyufWY6nk6FDB5MJCV00L7/82j0ja7Vr17zZV19tuxAWFl5xe1nfvl0bjR07IetxJL6rPtaqVct9CgryxDNnzs1+1OPWxXffbVXv3//DY0+KdubMSdc5cz4J3bv3QPrjHPdBuJ9nNXLk0MgXXkjUDhgwiAv/Wgf1/XVcgX/hyJ0kyY4APsEtDQUATAJQQFEUQ5LkYAA/kSQZViWM1Id/Eo6VrrACAAwa7VMfxvNZCFX6LKyRZWhIS9JRsnup8x4hdQMIHgi5CrJeH6GMJwa01pul/+x5GCrKsDRtLQAWReZK+V9pkcOvXImWtluaBYIg4OauQIsWbeDnF1Cjjwd9F9XfJ8OyKLHacMVgrgyJyjCgWRal1lumR4FyCWK8FfCUiCAV8BDpLoOEzwcAmPQWmGod5d/lCQnHysHxzDJixOjCqv/XJ/PzpEnjQ06dOqESCATODYeXl0/Fli07Lj6G6d43I0cOjbx8+aILADgcDgIAqubeoUMnTVJSV11d7R+EJ+FZrVjxudPf9t8Szv4L1EtwoChq+kMaLweAP0mS/JvaBj4Av5v3a0CSZBsAmwH0oSjK+eIoisqr9v+vSJJcjMrcEv/qCQAHx78NyzIgCB5ozQ3YLx+H/e/DMFYrl3R8G0Ky/QONwbAMMnTXUGAqwv6sQzDZK6Mt8RgCHmZXNDQFQuwQQMBWbsJFIhGCgsIgkUjQuHEc7vYD+yCwLIurOiMuFOpAGUy4WmZxlkn5PHhKROARQJinFL4yCUh3GRTix+c7wfFsQ9M0+DeFUo5nj759BxQmJ4/P/7fnAVR+V7Isi7slDay+cZ40aXyIh4eXrfrcz5w5+UhVnk/Ss+K4O49VH09RVDFJkn8AeAWVQsErANKq+zcAAEmSLQBsAzCAoqjfbyvzrxIeSJLsgsrIS3l4XNQzYR4Hx+OCKSuGedcnYK01T6MJuQry8Mag/ZqCH9j4vmz0aYaG0W4Gw9LILLsBIU+AfGMhjuacQrn9pjjCAu4VUqjNrvA01/SJ8PDwQrt2L8DNzf2B1wcA2eUWnC7Sw0XIh79MDK3NgXRNOUwOGlaaqVG3nbcCciEfYa4yBMjrl1yNg6Nv366NevToU3zkyEF1UVGRuG3b9trRo5PzZsyYHEpRl1zCwyNNc+cuuqZUKulx40aHXbp00dVutxFBQSGW8eM/zo6OjrUClRsukUjMFBcXiS5duug6ffqnV+12O2/NmpQArVYjkkikdN++Lxa9887worrms3btSu/du7/3Jgjg9df/l5+Ssjj4q6+2XTAY9PxJk8ZH7t178M8qgWT//h8UX3213m/Llh0Xly1b5JednSkRCkXsuXO/KDw8PCtmzZp/7dChVOXu3d97C4UCdty4CVkdO8aX1TZuu3bNmw0bNurGrl3bvQ0GvbBnz75Fffr018ycOTk0J+eGtEmTOMPs2Z9likQiFgC2bNns8d13W3xMJqOAJGOMH388NdvX188OAMePH3VLSVkcqNfrhB07xmtv//n89ttv1N99t9XHYNALw8MjTRMnTs0ODAy23TmrO7l48YI8JWVxkF6vE7Zs2Vo/efLMbIlEwup0Ov7UqRNCr17NkNM0TZBkjHHChCnZ/v4BdqDSXKpBg8blf/6Z5paTky2NjCSNs2bNy1SrPRwA8P3336o2bvzS32q18vv06V9Yfcxlyxb55efniefOXZg5atQwEgC6dOnYFADmzVuc0bx5y3orK/fu3an84ou1AV99te1vNzc35siRg26ffTY39Kuvtv7t4eHpaNeuebN33hmes3v3994Wi5kXH5+k+eCDj3Nr2/T/9tsv8mXLFgUVFuaLfXz8KsaMGXujRYtWpqr1xsY2NKan/+malXVd9sUXm/+maQexcOG8wMzMa3JXVzfHm2++nderV996axPWrVvtvWPHtz48Ho8dMuTtvJdeekUDABUVFcSSJZ/5nz59Qulw2HmtWrXRffjh5BypVMpWmTn16dO/qLa2T8qzSkjooomLa2ZcsWJpME07iISEtk15PD578OCJP+r7fJ526i043NQAvA0gCndmjgZFUS3r2dVwABtJkpwKQIdKHwaQJLkPwFSKos4BWAlACmANSZJV7d6gKOqvm229ATAAygD0pijKUd91cHA8LbAsA9uv22H7c98dZaJm/SCK6w0vL7c6zVscjAMECPB5fPxenI4vLmyuc8wAoQ/iCBI07YA9v+ZvZKNGTRAWFgV3d8U/W9BNjHYHKIMZlN4Ek4NGicUGo+PulogiHoEXQrygIogaJkccTza6I4fUZWdOezzKMdzSCuNvAAAgAElEQVSeb1uqjE+st73yqVMnlEuXrspwOBzE22+/EZucPEo2YcLkrMhI0jpmzPDIzZs3eI0enVzQqlUbw/Tps7NEIhG7cOG8gJkzp4R9880tk4pTp46rPv10wZXmzVtetdlsxIsv9mw0ZcrM661bP2/U6XT8nJxscV3zOHr0kNuuXTt8Fi1aTgUFhdhmzpzidAJq2rSZ2cXF1XH8+BG3+PikMgBITd2nTkhIcq7z/PlziunTZ12dNWte5pQpE0I++GBMVOfO3Ur27ElN3759m3rx4gXBHTvG/3W38c+d+8Vt/fpvLubn54qGD38r9tKlCy7Tpn16XalU0UOHDo7es2enasCAQZpTp064bty4zn/+/CVXSDLGsmDB7IDJkz8K++KLTZRGUyr45JMp4cnJH2Z17txNv2nTes8DB/Z7JiZ20dycs2Lr1q99585deDUsLMK6Zs0K36lTPw5bv/6by/V5V0ePHlIvWpSSIZfLmbFjR0euWrXcNzl5fD7LMujWrWdpp04J12maxvTpE0MWLJgdtGTJymtVbY8fP6JasGDpFX//ANuYMcOi1q9f5/3BBxPyKOqSZMWKpcGzZs2/EhfX3LR48Xx/rVZba8r5lJQ11KuvDmiUmno87Z9oUnv16qc7deqkYt68T4LGjZuQs2jRvJCxYz/M8vDwdO5nTp8+qfjyy80XTSYj///+772ob7/dYr3dP0Sn0/InTRofOXz4qBu9e/fX/vjjHtWkSeMjt27d+ZdKpaYrn9Vh1fz5i69ERERZTSYT/403Xmrwxhv/y+vXb+2VS5f+lo4f/39RUVGkhSRjrLfP83YMBr3QZDLy9+xJTT9x4pjbp59OC0tM7KpXKpX0okXzAgoK8sUbN265KBAI2UmTxoetXLnUb9y4CXn3avukPCsAiIqKto4c+X/ZnKlS7dQ3c3QSgH0ADgNoB2A/Kjf2bVEZ6ajekZYoiroMoFUt97tX+3+LOton1neshwl3aMnxJMAYNbAe/xJMWTHY8luKOvHzr0HYILHep+t/lV7E6vQNddZp7t0E17SZiDYGwkvhCavBDH2uBlYYnHUiIqLh5+cPX98AiMV17oXuoIJmUGKxIcdkRZ7JiuvlFtgYBmbHLQ2CWiyEgEdAJRaiV5An5EI+Siw28HkEgl0qE64Bz4bPCsejp2/fF4u9vLwdABATE2t0d1fYGzV6zgIAbdu216elnXcFgIEDX3Vu0t97b0x+z56JTQwGA9/d3Z0GgObNW+lbtmxtAgCJRMLy+Xz2+vWrktjYhmalUkkrlUrznaPf4vDhg6qEhKTSKi3Gu+++l3/q1HFVVXlCQmdNaup+dXx8UplOp+Wnp//hNmHClBtV5dHRMeWdOiWUAUB8fKLu11/PKocNG1nI5/PRs2cf7YoVS4L1ej1foVDUumF77bU3C93c3Bg3t1hrQECgJS6uRVlISKgNAJo1a2HIyKBkADSpqftUCQmdNY0bNzEDwPvvj8/r0SOhyY0bWaLffvvV1d8/0NKjR28dAPzvf0OLd+7c7gzrvmfPTs9Bg14tjIqKtgLAiBGjCxITt/nk5GSL6qN16N27X3FAQKAdAF5/fUhBSsriwOTk8fkqlZru3r2XvqrekCFDC8aNG0VWb5uY2EUTERFZAQAdOsRrz549pQCAgwdTlXFxLQxt2rQ1AsDo0WPz9+//oT6BX+7K7t3f++zbt9fZR4sWrfSzZy/IAoCPP556Y/DgQbHvvfcO2bx5K0NiYhdD9bavvvpGoVKpopVKFd2374Cio0cPqm7fDB89esjdx8e3on//gVoA6NOnv3bnzu+8Dh8+qHjppZc1AJCQkKSp+iydOHHEzdPTq6LqpL9Ro+csrVu31R848JOSJGMK7rUePp/Pjhz5fr5AIEBCQpLhs8/mMNevX5HExbUwHTyY6vHFF5suKpUqGgAGD36rYNasaWFVgsPd2jZrVqmpeRKeFce9qa+YPBPAUgAfoTIh2xSKon4nSTIYQCqAY49mehwczzYswwCOCtClWbD8MO+OckmHtyAg299TYGBZFn+WXMDnFzbVuK8UK+AqkkNXYYCbyBVjGr8Dc5kJ169fQXl+GfgFnmBgRWFJpRtSeHgUQkLCQRAEfH3978sMiLnpoPynphxpmjLobTUVhVI+DxaagVIkgI9MjBdDvSET3Kk9CJA/+lCtHI8eZXyi5n60AY8DtdrD6UEvEokZpVLl/JCKxRLGYrHwaZrG4sUL/E+fPqEsLy8TEgSPBQCtViOoEhy8vLzs1fudMWP2tQ0b1vlu2LAuICgoxDJ8+KjcusxatFqNkCRjnOX+/gE1NtK9evXVDB48qKHJZOLt27dXRZIxRh8fX+eYCoWyxrxdXV0dVWZNUqmUAQCTyci7m+BQ/SRXJBIzKpW6xnPR6TTCm/MURUWRzk26i4sLI5e70IWFhcLS0mKhh4eHc94EQUCtVjuvS0qKRWvXrgxct26NM1ICy7JEYWGhsD6Cg7e3r7OOv79/hV6vEwGA2WzmzZ//aWBa2nk3k8koAACr1cqr7mtS/T1LJBLGarXyAECjKRF6eno6+5XL5YyLi+sDWTT06dP/rnb7CoWCbtu2vW7v3l3es2bNv3Z7ua+vn636/7Va7R3OWSUlJSJPT68akac8Pb1tJSXFzrpeXj7OfgoKCkRXr16RJyV1aFJ1j2FookOHF+r1t+ji4uqormERi0WMyWTmlZaWCGy2Ct677w6JuVWbBcMwxL3aVl0/Cc+K497UV3CIBTAZleZBLAA5AFAUlU2S5HQAMwB89SgmyMHxLEIXX0fFue9rzeIs7vA/iKI73rMPlmVR4bDhx+sHsC/rUI0yhdgdHzQbCYXYHQRBwOFw4JtvvsT3l7fc0U9wcBjc3RWIjW0Ekaj+WgUbzaDYYkOJ1YZ9OaUwO+g7QrMl+Kmgq7DjeW8F/DiBgOM/wO7dO1S//HJGsWTJyoygoGCbwWDg9+yZ2IStYcBP1PioN23azNy0abNrdrud2LRpvefMmZPD9+y5e0hLlUplLykpcprI5OXl1jCX8fPzt0dGksbU1B8Vhw4dUPfu3bf4oS3wPlCp1LbCwgLnl4LJZOKZTEa+j4+PXa32tJeWnnbOm2VZaDQa57Va7WF75ZU3Cvr1G6D9J2MXFRU4+8rPzxMpFEobAKxfv9Y7Ly9Hsnbthkve3j6Ov/5Kl44Y8VYsWw//RJXKw37jRra06tpsNvOMxvJa90kPw3fqr7/SpYcPH/Bo27a9dtGieUErV667Ur28oCBfVKUpKCzMF6lUKvvtfXh6etpOnz5RI0NnaWmxqGXL1s4T+epz9fb2scfENChfvfrLGmM9KGq1h0MkEjEbN27528/P/455PiiP61lVwfnG3Z36Cg5WADyKoliSJAsAhAM4ebOsDJVRjZ4JWJa5dyUOjn+I9czXsF84eOuGQARBcFOwRi1Ecb0hCGx017bF5hLkm4pwUUNBxBPiaO6pGuUxqigMazwEQp4AJpMRe3dth81258GeSqWGr28AGjRoDIfDAReXewfS0FfYkWOyosRqh9HugMZqx5WymtYYEj4PEW4ydPBVcloDjv8sJpOZLxQKWaVS5TCbzbxlyxb611XfZrMRP/64Rxkfn2Rwd3en5XI5U6WluBvx8UnahQvnhfTs2UcTEBBkW7dute/tdTp37qbZtu1rn9LSEnGXLj30tfXzqElK6qqdM2dGWNeuPTVRUaR16dLP/MPDI01BQSE2udzFsHr18qD9+39QJCV11W/atMHLYNA7T3Z79+5fsmHD5/4xMbHm6OhYq8Fg4J88ecytZ88+9XLS3bt3l2enTgl6mUzGbN680bd9+446ADCbzXyRSMy4uytonU7L/+KL1X71X08X3ciRQ31++eWsS9OmzUzLly/2Y1m21h2kWu3pIAgCWVmZ4iqzp/vBarUSs2ZNDRs8+K28l156pfTNN1+O2bRpg+cbbwxx2qBu3brZp0mTZiaTycjbvft7r/79X7rDob5TpwTDqlUpQbt27VD17NlHu3//XmVubq4kPj7xjs0wAMTHJ+m//HKt//fff6vq1aufDgAuXEiXyuVypsps7J/A4/GQmNildNGieYEffTT5hqenlyM/P0+YkXFZWmU290/5N56VWq22a7WlIpvNRlQFAuCopL6Cw58ASFQmgjsM4GOSJPMA2FBpxnRXJysODo67w9IOWA+tAGszgy645YPFD2wEafxwEGL5Pfu4UHoJq9LX11rmKnZBuGsIOri0RPbVq9iy+csa5TweH2q1BxiGQWxsI4SGRtzRh9lBw+KgIeDxkG+yosRqg5DHg9lB3xQWbNBV3KnND3OVwl8ugb9cDLVYCH9OWOB4CujXb4Dm3Llf3Pr37/6cXO7iGDz4rfwDB/Z71tXm4MGf1KtWLQtiGIbw9fW3Tpw4LbOu+vHxSWUZGVTx+++PJHk8gn355dcLTpw4qhaJRM6Tqy5duutXrFgS3LJlG51MJvvHJ1ozZkwKAoBp0z69ca+6t9OhQ6fynJzs/KlTJ4SbzSZBVFS08ZNP5l4HKk+gJ02afi0lZUnQokXzQjp2jNdGRUU7I0R369ZDb7GYeNOnTworLS0Ry2QyunHjpmW1CQ6rVi33uXAh3bV6uNBOnRK0yckjo3Q6nbBFi1b64cNHFwDA668PKZo6dUJYjx4JTRQKpf3FFwcWnjv3a70iNkRHx1pHjBhz49NPp4dWVFTw+/TpX6hSqWo1m5LJZMyAAYMKRo9+N9rhoIm5cz+7Yrc7iEmTxkcePnw6rarerl3bfX74YZd31bVQKGR++unYn4sXzw/w8PC0vf565eZ38uSZmePGjSKff75dWXh4RAUAtGnTTv/WW6/Fms1mfnx8YunAga/ekThPpVLTs2bNu7p8+aLAFSuWBHl7+1TMmjXvalWUqNtxdXVlPvtsWcbSpZ8Ffv75qkCGYYng4BDz6NFj7wiLf7+MHftR7sqVS/3efXdITHl5mUCpVNl69OhdgsoD5nvyJD2rtm07lO/Y8a2lV6+k5wiCYH/66dif//jBPGUQ9VHfkSTZHUAoRVErSJL0B7AXQJV9XC6AfhRFnX9003xohADI/CcJ4A6u+gBaiw0dXugF36ZJ927wH+ZZcDT9N9fIsizsFw/DfuUMmOLrNcr4fjGQdnkfhPDuJkFGmwm/F6eDz+Nh7/VUlNtuZWto69cSIp4IDT1iEOYeDHcXMXbs2InS0mIwBA8AAR5LgwWBmE7dYRRI4CMVI1ohR2a5BQabA3IBHwzL4nq5BX9py1Fmrzu3oquQD3+5BC093UGzLEQ8AmqxCCrJ4zMb5T6zj5ZqCeBCAWT9037S0v7ICggIv2fWYI6aZGRclrz99hsNjh49WyPZWL9+3RsmJ3+Y/TgyKHM8XurKnM1RE+5ZPXxyc695NG3aJKS2svomgNtX7f95JEk2AxCByshKlymKqlfc5f82nL0bx4PBmPVgbWaYd38KVNzyixS3GwxBUBMQcgUIovbEPBW0DTN/XgB9Ra3aZyTHjUCEItR5XV5ehvU7dsAukqLM1RNmMhoMr9IxUMznoYJmQBWZAdQZ3AUAQLrLEOEmQ7HVhgC5BCqxEAzLwl0khJjPg5uQz9mDPsHQRiMMJ4/DmpUJjxcHQuT1QEFiOB4T+/f/qIiPTzSYzWZeSsqSgGbNWuirCw379u1VEASBdu06cEIDBwfHY+O+gw+TJEkA8AWQ+UzmT+ASwHHcA5a2g+ALwbIMrMe/gCPj9B11CKkbZL0ngufuU0sPlRSaivFT1hEYKgzI0NcMIhHkGoCmXo3Qzq81ZMJKXz673Y7T537Gn6VlKHf1hDm0eY02DZUuqKAZZBktzutQVykyDCZYaAY3jFb0DPKEXMCHiE9AwufDXSh4rJoDjocDY7XA+PvvMP7xO0x/XwBbUQFRQCDA1K094nh8rFq13Gf79q13+C5ER8caV6z4/MrevTs9Fy2aG8Lj8RAT06C8erjVoUMHk7m5uZKPPpqUebcswBwcHByPgvtJANcdwDRUmigJALQA8DtJkp8DOE5RVN2Zozg4nnKspzfD/vehOuvwAxpC3Gog+Oqgu9ax0XbM+W0xis01LTr6hndHUnCnO+obTCas/iMDBpELIPABfCqFERmfBx9XKZQCPnoGeULMv7XBYFnWqSVo4/1gCds4nhysN7JRdvYMyk6dAGOxgBCKUBHZENd8G0Hr7oN+LirUms2K47EzYsTowhEjRhferfz2qDHV+fzzr7ikVE85p06d+y+Yfz8RcM/q8VLfBHCDAXwJ4GtUZnWu7omZgcqM0pzgwPHMYc84BZZ2AIyjVqGBELtA1ncKCLkShKDuLZvVYcWyPz5HdtktH7UBkb3RMeB5AADvNjMmk53Gqj8oaCEERC7O+x18FGikcoW/XHJXu3jOtOjpgXU4YEz7HYZTJ2C+eBFgGRS7++G8dztcl/qg3EZAmEdAVW5AmckGhcv9Jerj4ODg4OCoor4ah0kAFlAU9TFJknzUFBz+BvDBQ58ZB8cTjP3qz7AeWX3HffnrS8CT1f8En2ZobLy4FeeLawZs6BnaBV1CXrhDWNDptDj268/IpAUo9QwFUGlG5Ak7+keHIshFwgkFTzksTcNyJQNlV66iLD0dbN4NELYKmIQy/O0ejbPKhnBRKxDm5444ER9NIz3RMFQFHo/7XHBwcHBwPBj1FRyCURmKtTasANweznT+A3A+Ds8sLMsCNjMsR9aAzrmVu4nnFQ5CKIGoQWK9hAaaoZFWnI5d1/ZDV1Ez/Hpr3+Z4LXpADYHBZrMhvygfWy/noEzhA6giAQBCmwVNUY4eLVtCKOT8EJ52Mk6eQ/bZ38FP+xlic6WTvJknxg2pNy6rwmAMikJcjA8+beoPd7mIEyA5ODg4OB469RUccgA0BXCklrLmAK4+tBk9oXA/wc82rM0C44YRNe5JOr4NIdm+fu1ZFsdyT6PIXIKTeWdrlL1M9sfzvi3AI3g1Nnvl5WX4+co1/Ka3wCpxg0Nxy5H6lTAfNFC5gMdtDp9YWIcD4PNBG/TQHzmM8t9+gbzRc3Br1x6SoOB69ZGXfgk3vt8NaWEWZA4rZAC0QldcJl+APDoaqtBAhMsl6OLvDrGI/2gXxMHBwcHxzFNfweELANNIkiwCsOvmPYIkyQQAH6IyCdwzAqdxeBZgGQaOaz/DenTtHWWETAFJx7cgCGx8z35ulOVi4fkVcLA1o9kkBnVEM+/nEORae9L1sxlXsdfAAnABXCv9F9oqJegeHsCdJD/hVOTkoGT7Npj/vlDjvsg/AIYTx6A/cgjikFAoOr0A1xatwBNX+hzQZhOK0y6gvMyEkp9/gbQoBy4OM9wIPkp9IsBGRUDevBWi/L3Q2o1LpsfBwcHB8fipr+AwD0AggI0AqnZAZwDwAayhKGrZI5gbB8e/AsuyMO+YCkaXW+O+KK4PxM371dnWTttBszTEfDFO5v2MbRk7a5R3C0lAx4C2cK3mzFydQ1ez8avGBCOv0vRIAgYDI/wQ6iavERWJ48mBZVkYz/8Gw8kTsBcXwV5SAkIsgVvb9uC7uoInlUJGRkMaEQnaaETZL2dhOH4URRu+RMGWb0B7BYDWlEBivpWjwxNAibsfLOGtENQ1EQ3C/J+JJHccTzdDhw4mExK6aF5++bV7JgGsK6lX375dG40dOyHrcSS+qz7WqlXLfQoK8sQzZ87NfpA+R44cGvnCC4naAQMGaR7WPIGHkwht0qTxIR4eXrbk5PH5P/98xuWzz+aEbN++98K9Wz5ZcEnhHh31TQDHAhhJkuQiAAkAPABoARyhKCrjEc7viYPljJaeahijFqZvxjqvRU17QdSkJ8AXgODVbQpytuAcNl/69o77SUGd0Ce8W62agjKLFbvPnUcBK4JeetM/gieEiLbh7XAvBHp6PtiCOB4ZLMvCcvkSSr7dgoqcHBAiESRh4ZBGRUPdtz+ESiUAwGx1IKfEiKs/Z0NTZoW+3Bv5/r0g4WWhsZ6CX2ERSkUK0OGxcIsIh6vCBTHPN0GUnNMqcHA8SVQPn3vjRpbo1VcHNDp27Oca2byrM2nS+JBTp06oBAKB01Th/fc/yFqx4vO7htp9WFQfWyAQsKGh4aZx4z7KiYwkrfXto3Xr540PKjScOXPS9aOPxkaJRCKGIAgoFEr7oEGvFrz00isPVWh6mNz+bqsLU//23J4E7isBHEVR1wBcu2fFpxHnno8zVXraYFkWjDYHtvO74ci6FQ7a5Z11IHj3/hMxVJThRnmuU2gQEHw4WBo+Mi8MazwEXjKPGvX/LNLgSF4pLHYHjDwRILmVyZdgaLwdokSYt/dDWh3Hw4Y2maBL3Q/DmVOg9XrwXd2g7tMPqm49QAgEyCs1Ye3R67iQlQ6HgwHN3PrO4BEEVG5ieLhL4OLVEI1adIdUIkAjFzFkkvvOx8nxH4CmafD5nP/Js0rfvgMK/60NZ9XYZrOZN2PG5OBZs6aHbNy45fLjnodCobTv3XsgnWVZpKbuU8yePSO8ceMmJpKMqbcQ87hwOJ69vMb3y11/qUiSvHuGqlqgKOrGvWv9h+HkhacO1m6F5dAK0Dl/1bgvCImD5IV37yk0FJlLMPPnBTXuTWs9Hl6yW1qCCpqBlaahsdpxUWdEWrEOehoAeABPBLHViFgJEBQcijBXGTylXHquJxXaaIT+6GHojx0BbTBAEh4B987dcFVN4mSxBZlf/Y4ysw0Gow0SER8tor3gKhNCIuRD7S5BgxAVXOUizqH9CaVv366NevToU3zkyEF1UVGRuG3b9trRo5PzZsyYHEpRl1zCwyNNc+cuuqZUKulx40aHXbp00dVutxFBQSGW8eM/zo6OjrUClSe9IpGYKS4uEl26dNF1+vRPr9rtdt6aNSkBWq1GJJFI6b59Xyx6553hRXXNZ+3ald67d3/vTRDA66//Lz8lZXHwV19tu2Aw6PmTJo2P3Lv34J9VAsn+/T8ovvpqvd+WLTsuLlu2yC87O1MiFIrYc+d+UXh4eFbMmjX/2qFDqcrdu7/3FgoF7LhxE7I6dowvq23cdu2aNxs2bNSNXbu2exsMemHPnn2L+vTpr5k5c3JoTs4NaZMmcYbZsz/LFIlELABs2bLZ47vvtviYTEYBScYYP/54aravr58dAI4fP+qWkrI4UK/XCTt2jNfeHpTw22+/UX/33VYfg0EvDA+PNE2cODU7MDDYVp/3dfHiBXlKyuIgvV4nbNmytX7y5JnZEomE1el0/KlTJ4RevZohp2maIMkY44QJU7L9/QPsQKW5VIMGjcv//DPNLScnWxoZSRpnzZqXqVZ7OADg+++/VW3c+KW/1Wrl9+nTv0aCvmXLFvnl5+eJ585dmDlq1DASALp06dgUAObNW5zRvHlLU33mXt1ka8aMyUEGg164aFHKNQD47LM5/leuUPLVq9dn2Gw2YsmSz/xPnz6hdDjsvFat2ug+/HByjlQqZQHg889Xee/atcObIIDBg9/Ku9t4MpmMSUrqqp0zZ0YYAGRkXJYsWDA7OCsrS6pSqexvvz0st3Pnbobb2505c9J1zpxPQvfuPZAOALm5OcKFC+cGXb580YVhWKJ9+47aDz+clNOrV9JzixevoGJjG1oAoKSkWDBoUN9G3367u8aPK0EQ6Nq1h37p0s8cV69ekYaEhFUsXDgv4MyZE0oAeP75Drpx4z7KFYvFbNXY3bv3Kt69e4e3RCJh3nzznbx+/QZob3+GAPDdd1vV+/f/4PHll5vvSI546NAB9/Xr1/oXFRWKZTIZnZjYtXTMmLH5wC3twujRydlff/2Vr6enl23atE8yq9pu2bLJ49SpEyoA+OGHXd4NGjQqX7Zs9dWCgnzhggWzgy5d+ttFLJYwffu+WDRkyDvFVZ+TB/kbfNKpa2eUWUdZdQhUbqufiSMV7if/vw9j1MBakQfjxkk17hMyBeQD54AQSe/Zx8gjH9a4jlWRiFa3xZK/9WjhyaClpzv+0JTjdJG+1vbRJRkIlIvRqkUbyGTyf74YjkeO5eoVGE6dQNmpkwAAUXgkihIG4rrAA+epEhhM1yES8BDo5YJgb1eoIsTo0y4U7lyitTr563yeOuNCkce9a/5zohp6lzZq5l9vk4hTp04oly5dleFwOIi3334jNjl5lGzChMlZkZGkdcyY4ZGbN2/wGj06uaBVqzaG6dNnZ4lEInbhwnkBM2dOCfvmmx0Xb/VzXPXppwuuNG/e8qrNZiNefLFnoylTZl5v3fp5o06n4+fkZNf54Th69JDbrl07fBYtWk4FBYXYZs6c4gzD1bRpM7OLi6vj+PEjbvHxSWUAkJq6T52QkORc5/nz5xTTp8+6OmvWvMwpUyaEfPDBmKjOnbuV7NmTmr59+zb14sULgjt2jP+rtrEB4Ny5X9zWr//mYn5+rmj48LdiL1264DJt2qfXlUoVPXTo4Og9e3aqBgwYpDl16oTrxo3r/OfPX3KFJGMsCxbMDpg8+aOwL77YRGk0pYJPPpkSnpz8YVbnzt30mzat9zxwYL9nYmIXzc05K7Zu/dp37tyFV8PCIqxr1qzwnTr147D167+p14n40aOH1IsWpWTI5XJm7NjRkatWLfdNTh6fz7IMunXrWdqpU8J1mqYxffrEkAULZgctWbLSaS1x/PgR1YIFS6/4+wfYxowZFrV+/TrvDz6YkEdRlyQrViwNnjVr/pW4uOamxYvn+2u12lpPc1JS1lCvvjqgUWrq8bS7mSrVh/HjJ+a+8cbA2O++26oODAyqOHQo1eOLLzZfJAgCixbNCygoyBdv3LjlokAgZCdNGh+2cuVSv3HjJuQdPXrIbefO73wWLqz6jEy+a6g2o9HIO3hwvyokJNRst9uJCRPGRSQldS1dsWJdxq+//uwyffrEiNDQ8IuRkVF39QegaRrjx78f2bjxc+WzZs3/i8/ns3/+mSYXiURs2+q5GZUAACAASURBVLYdtPv27VXFxjbMA4AfftijatiwcbmHh6cjI+PW62QYBqmp+xVms5kfFRVtXr16uS9FXZKvX//NRYIgMH78/0WsWrXc9/33P8gHAINBLzQY9ILdu1PTf//9N/mkSR9FNmzY2FTXPGtDJpMykyZNz4yOjrVcunRROn78mKioKNLctWsP5w/0H3+kuWzZ8v3fPB6PLSkpcsY3f+WVN0ovXEh3qW6qxDAMxo//v4jWrdvq585ddD0/P0+YnDySDAkJtXbqlFAGPPjf4JNMXZ92AoARwG4AewCYH8aAJElGodLJWg1AA2AwRVFXbqszBcDLqHTEtgOYSFFU6s0yGSoT0DUD4ADwAUVRPzyMudULTvPwn8X68zbY0/cDAKofC8lfWwyeXHnP9rN+WYgCU81Dwu5hryLILRwGmwP7cir9/X4rKcNvJTUPEngOO1S6XLjbjEhq3hxBLXo82GI4Hjnmy5dQuncPrNQlAECOIhjnfeJw2eEKpFdAwM9HiI8bBsZHoEmEB6RiztTov07fvi8We3l5OwAgJibW6O6usDdq9JwFANq2ba9PSzvvCgADB77q3KS/996Y/J49E5sYDAa+u7s7DQDNm7fSt2zZ2gQAEomE5fP57PXrVyWxsQ3NSqWSViqVdf6eHj58UJWQkFRapcV499338k+dOq6qKk9I6KxJTd2vjo9PKtPptPz09D/cJkyY4tT6R0fHlFdtYOLjE3W//npWOWzYyEI+n4+ePftoV6xYEqzX6/kKhYK+c3TgtdfeLHRzc2Pc3GKtAQGBlri4FmUhIaE2AGjWrIUhI4OSAdCkpu5TJSR01jRu3MQMAO+/Pz6vR4+EJjduZIl+++1XV3//QEuPHr11APC//w0t3rlzuzOm9J49Oz0HDXq1MCoq2goAI0aMLkhM3OaTk5Mtqo/WoXfvfsUBAYF2AHj99SEFKSmLA5OTx+erVGq6e/dezg3hkCFDC8aNG0VWb5uY2EUTERFZAQAdOsRrz549pQCAgwdTlXFxLQxt2rQ1AsDo0WPz9+//wQsPwO7d3/vs27fXCwD4fB7700/HamT7lMlkzMSJ0zInTvwgUiKR0iNGjMnx9w+wsyyLgwdTPb74YtNFpVJFA8DgwW8VzJo1LWzcuAl5hw8fVMXHdy6NiWlgBYChQ9/LP336pKq2sYVCIRsREWmaPHlm1vnzv8mtVit/2LCRhTweD+3adSiPi2uu37//B3Vk5Ni7mlSlpZ2X6/Va4fjxE3OqBKVWrdoYAaBHj96aGTMmh40bNyGPIAgcOXJAPXDgq05tjV6vEyYldWjC4xFQqz1s48ZNyIyMjKo4duyIatSo5BxPTy8HALz55tv5S5YsCK4SHKregVgsZtu0aWeMi2tmOHBgnyoyMqrgft7B88+3dzrRN2jQ0NKuXSdtWtp51+qCw7BhI/PlcjlTn/7S0s7Ly8rKBKNGvV8AACEhobYuXbqXHDqUqqr6u3vQv8Enmbp+6doDeAXAAAB9AewFsBXAfoqi7A8w5moAKyiK2kyS5OsA1gCIv63OrwAWUhRlJknyOQDHSZL0pSjKgsos1WUURUWQJBkJ4CRJkhEURRkfYE735qaqgZMb/nuwLIOK05thv1gzDYmwQSIkbV+vsy3N0LAxdnxwYmqN+wK+P+Sy7jhdApwuuaXN9s37G1JLGcrcfWCRuMI/72/wWAYdOyYiuE3nh7cojlphHQ6Yqcsw/ZkGEDy4d+gIsX/tIW/vaMuyyC8uRwF1HcLj+yHNpsCAwEXXMPyqiIUyIgwCPoEkDxe0ivVGiI8rl435AWjUzF9zP9qAx4Fa7eH8bROJxIxSqXIaPIvFEsZisfBpmsbixQv8T58+oSwvLxMSBI8FAK1WI6gSHLy8vGr8Rs6YMfvahg3rfDdsWBcQFBRiGT58VG5dZi1arUZIkjHOcn//gBob6V69+moGDx7U0GQy8fbt26siyRijj4+vc0yFQllj3v/P3nnHNXH+cfyThCzCyGAHkCEcoCKIe2FB3Fur1lo7rKtVf9XW1oqrVq17r1qts+7WWfeo1lG3deHhYu8sSELIut8fQCAyDI5W7L1fL17knnvmXXL3fJ/nOxwdHY2lak18Pt8MABqNmlnVpMXFxdVSnsPhmsViidV1UShk7JJ+coKDCcvky8HBwSwQOJiysrLYeXk5bBcXF0u/GQwGJBKJ5Tg3N4ezdu0qn3XrfrT8QCmKYmRlZbFtERzc3T0teaRSaZFSqeAAgFarZc6bN8vn5s3rThqN2g4AdDods7ytSfn7zOPxzDqdjgkAMlku29XV1VKvQCAwOzg4vpTSe8+efZ5r4xAV1UTj5uZepFKp7Lp16ykHgLy8XDu9vog5fPhHoWU5KZjNZgZQ+h0JsXxHfHx8K1yzytq+e/dve4nERc9klnnpc3Pz0Ofl5VYbRTQzM4Pj4uKqr2x3JSqqiYbL5ZovXvzT0c3N3ZCVlcWNi+tk+V6U2jg8W06pVHCkUqll90Aq9dYrFApLPwQCgbH8ZN7NzV2fl5dX42in169fFaxZs0KalpbCNxqNDIPByGzRoqWifJ5SVTZbyMhI5yiVCk5cXNuI0jSKMjMIItQioLzsb/BNpkrBgSTJCwAuEAQxFsWelAYC2AgABEHsQ7EQcZIkSZsktJJybgAaAYgrSdoOYAVBEK4kSeaWa/tYuWK3UTxtlwBIAzAAwIcl+R4SBHENQGcAu23tx8tBiw61AXN+DgBAs3MiQJV9RQUD54Hp5PZc15aFRh22P/gV13OsFofwbsg4/JmlRaGprM4Avh3yc7PAUWZBolPB01MKtTofPLMJzXsPgIOD4yseHU15zAYD1E+eQnXrHmS/H4Qxr8zTo/LUCbCcnODYtBmcW7UBx9ungnerzJx8/LX5NzgmP4CnLhdOlAlGBhPXXRqgqFFrREQGoru/GGw72h0uDbB//6/iy5cvCpcsWZXo61tHr1KpWN26tY+grBT4GVYvisjIKG1kZNRjg8HA2LJlg+uMGZMDDxyoOJEqRSwWG3Jzsy0qMunpaVbqMl5eUkNQEKE+dux34cmTxyU9evTKeWUDrAFisUSflZVpUbvSaDRMjUbN8vDwMEgkroa8vAuWflMUBZlMZjmWSFz07733QWapznpNyc7OtNSVkZHOEQpFegDYsGGte3p6Km/t2o0J7u4exjt3bvNHjfokjHrWwKLS8bgYUlKSLbqqWq2WqVYXVDpPepXxdLZs2eBqMBgYIpHY8NNPqz1GjhydJZG4GDkcjnnTpu33vLykFSa1YrHYkJ1d9h1JS0u1yUDO1dXdIJPlccxmM0qFh9zcbI5U6lOt+o+np5c+Ly+PYzQaUZnwEBvbQXb06GGxWCw2tmjRUsHj8Z57wYVCkT49PZ1burOWkZHOEYlElrFqNBo7jUbDLBUecnNzOH5+/oUAwOXyTDpdoeWhLJdXLVDMnDnNv3v3XjmDBv34kMfjUbNnz/DJz1dZDYLBYFTZ32fvtaenl97V1a3o118P1To3ta+C5+6tlwgGJwCcIAhiJIBOAAYB+B3AHhTvStiKD4B0kiRNJXWbCILIKEnPraLMEACPSZIsdarvC6C8D+WUkvKvGXplsTZgyn4E7f6ZlZ7jd/kKTKfn7zpfz76Fn+9ts0pz5vkC7I44nl62seXPKALn/kUwUOyfGADeHfRJpQ9VmleLUaVCztbNKHz6GCZlmR0Jy9ERbh98BIeISDBYLORfvABtwj0oz5yG8uQJMO0FsAtvhAcsV+TLVeClPYabToZ6hgKYmCzofINhFxwEhwYNMSA0gDZkpqmARqNlsdlsSiQSG7VaLXPZsoXS6vLr9XrG778fEMXExKmcnZ1NAoHAXLpLURUxMXHyhQvn+nXr1lPm7e2rX7dujeezeTp06CzbufMXj7y8XG7Hjl0rN6Z6zZQa3Hbq1E0WHEzoli5dIA0MDNL4+vrpBQIH1Zo1y32PHDkkjIvrpNyyZaObSqW0TO569OiTu3HjT9LQ0DBtSEiYTqVSsf788w+nbt16Kqprs5SDB/e5tmsXq7S3tzdv3brJs02baAUAaLVaFofDNTs7C00KhZy1fv0aL9vH01Hx+efDPC5fvuQQGRmlWb58sRdFUZU+BCQSVyODwUBS0lNuqdrTi/Do0UPuli0bpIsWrST5fL551KhPQlu1aqtq0CC8sH37jnmLFs31+eabySmurm7GjIx0dmLiA367drH5MTFx8gUL5vh3795T5uNTR//TT6ttGmejRo01XC7X/NNPqzw++WRE9tWrfzlcv35V+OmnIxOqKxcZGaURCkWGhQvneo8e/UUGi8Wibt++ZV+qktetW0/Z0KGD6/F4fNPEiVNsspFt2/Yd+datGzwbNozQMBgMbNr0s2d0dKzVLuSKFUu8xo2bkH7jxjXB9evXnIcP/ywDAAID6xZeuHBO1L//oLysrEz28eNHXZydhZXuGuh0OpaTk7OJx+NRN25cs//zzz/EDRtG2myYLBSKDVlZGRYBOTIySsPj8U0//rjSY8iQT7I5HA6VmPiAp9PpmJGRUa9Erf9NpqYznHAAbQG0QrH9QQXr9VcJQRDRAL5H2Q7FK0EiqTz4VnWUaiQ4CLhwdX37V5Br0xh1aQ+Q8YyhcylsVx94vjcVdo7iCueeHaPOWISfr+/EH0mXLGmdgj/FtWwmDOVcar4b5I6/D+4Cs2Qng8vlIjIyEuHh4XB0fPOuW226l1VhyM9H1tHjyL93H2a9HgVkIiiTCQ7BQXBq1xZsZ2ewuFy4xbQDi19m3O7h3w9APygzc3F93ymo7tyB+PJFBJSL5G0QusCl9wAE9+gIBvPN3lV4G+5lbad3736ya9cuO/Xp06WhQOBgHDLkk4zjx49UG3DlxImjktWrl/mazWaGp6dUN2nStGonVjExcfmJiWTOF198TjCZDGrgwMGZ586dkXA4HMtWZ8eOXZQrVy6p07RpC4W9vb3NO//P8t138b4AMG3arBp7Rmzbtl1BampyxtSpEwO1Wo1dcHCI+vvv5zwBAInExRgfP/3xihVLfBctmusXHR0jDw4Osay8dO7cVVlYqGFOnx4fkJeXy7W3tzeFh0fmVyY4rF693OPu3duO5eMftGsXKx837vNghULBbtKkmXLkyDGZADB48EfZU6dODOjaNTZCKBQZ+vbtn3Xt2hWhLeMJCQnTjRo1NmXWrOn+RUVFrJ49+2SJxeJK1abs7e3N/foNyBwzZniI0WhizJmz4KHBYGTEx08IOnXqwk1b2jMajZgxY3JA3779s+rXb1AIFHtHmjVrmv+mTTsSxo//Jm3VqqVew4d/FFpQkG8nEon1Xbv2yAWQHxMTl//wIZk9btxogslkUEOGfJJe3g6mKjgcDvXDDwseLlgwp86ePTs9RCKx4csvv336vPgOLBYL8+cvfrRgwRyfPn26hjMYoNq0aScvFRy8vX0Mfn4B2uzsTG7z5i1tUh0fNWpM5sKFc1gffvheGAC0bNlaMWrUGIv9grOz0ODo6Gjs0aNDOIfDNY8e/b/k0n4OGfJJ9uTJ39j36NGhoY9PncLo6Bj5rVs3Kn1Ajh79Rcratau816xZ7hsaWq+gVas2CrVabbNDn969++VNmfJNYFxc24h69eoXLFmy6vGCBUsfLV48z7tv327hRqOB4ekp1Q0dOqJKz1ZvE4znbd8RBBGGYjWl9wDUAXAaxWpKe0mSrOC+6zl1uQFIBCAp2W1godhAOqi8qlJJ3hYAdgHoSZLkjXLp9wB8SJLktZLjQwA2kSRpi6qSH4CnMpkaZnPNVI5OrvkaMq0Obdp2glfjLjUqW9uoLRFqzfk50N8+WsF2gdtqMDj12ldbtvwYzZQZky/MgkpfNuYR4WNwJN0Ilb5MvbWZgAHVNeu2+vR5741WRaot97IqTGo11DevI3f3Tpi1WrCcnUGZTBDUawBRXAfw/PwBlI2Toihoi4zIkmlx54kMD5IV0BvNSM/TwGA0g8EAWgQ4Ik6ohquPB/h1g8CoJTtE/+a9LDZqdAAAfwBJL1rPzZu3kry9A58bNZjGmsTEB7yhQz+od+bMJatgY717d6k/btzXyf9EBGUamucxefI3fhKJyysJlPasK1iaf5a0tMcukZERfpWdqy6OwyQUCwxhAM4DWAhgD0mSL/zQJ0kyhyCIWygWQraW/L9ZidDQBMBOAP3KCw0l7AYwAsC1EuPoJqiZuhRNLcacn4OiS9thTLZe0GHXjwOv5fs1qqvQWIif7mwBqXhkSWvk1haPCwnseFq2+PK/QAluXjmPnIQyI2ip1AcxMZ1eqZ4rTRlFqSnI2rAeRSnFWolMPh9eo/8Hh4hISx6ZSofsrAKk56mhNWQgKV2JhGQFFAVlWgNuIj4c+Gy0qu+BJiFuIOqIaPUjmlrBkSO/C2Ni2qu0Wi1zxYol3lFRTZTlhYbDhw8KGQwGWrduSwsNNP86KSlJnMuXLwnXrdt8//m5aWoz1S21zQRQgOIJfDqAAABfEwRRWV6KJMlvbGxzJIBNBEFMBaBAsQ0DCII4DGBqyU7CKgB8AD+Wa+8DkiTvAJgPYCNBEI9QrC41nCRJ+sH5lmKSp0O7Jx4MvjPYwa2g//uwdQY2D+ygVjUSGopMeih1+fjq3DRLmgO/I1h2vnhcWJbPLSsRTvk5OJZQpjbZpUtvuLhUq5VA8xJQJhOyt2xC/vlzAACHqMZwjGoCjZc/zqfrkHP0ARgMBh6lq5CaU3E3vI67I96JlELowEWwrxBuwufH5KCh+TdYvXq5x549OyrYLoSEhKlXrvzp4cGDe10XLZrjx2QyERpar6C8u9Vhw4YQaWlpvG++iX/KfMPV62jefpYsWeB18OBe9759B2TWqeNnUwA/mtpLlapKBEEkwXYXQhRJkgGvqE+vEz+8rKpSm07wakKrKv0TUEUaqDd9XiGdKfEBOywW7JDoGq34//bwEE6lnrNK43FbgMupb5UWLeIi6+JRizk8h8OBwWDAgAFDwOHUrqBeb8q9fB5GpQKqc2ehPHMapoJ8GKR+eBjZCQV8EQq0etxIzIOZosBiMsBiMeDizEdoHRHcRXzU8XBEA8IdSakKuIv4b+0uEK2qRENDQ0PzT/BCqkokSVZagIbmn6LoxgEAANO9LphObjA+vAi7wGbgx46qUT1myoz9j4/gVOp5sJhuAMMOJlMunB0/ssr3XVQgUp4+woULxUJDVFQzhITUt/j+pinGVFAAfW4Ocnduh1mng2NUYzhHt4Ods032h1ZQZjNUZ88gd89uUEU6mByF+NOnDf7i+AEPtAC0sOfaIcxfhEHtg6sUDBztOfAQ27/84GhoaGhoaGiqpHZYBb4BlE1V6DgOrxP9neMourwLMJcZJQt6Ti7+8M7wGtVVZNJjw71tuJN3HwyGAM6OQyvN91U9bxzaswXbE85Y0po1a41ivwA0pehSkpG7YxsKE62dqcnS0yA7sA8OUY0hfCcWfCKk2lV/ymiEPjsLeceOofD2TZjVamjY9tjj3RmZPFc48NkY3MYfrRt4gsVigEWrYtDQ0NDQ0LwR0IIDzRsBRVEoPPgDTFmJVuncZv1rXFdyfirmXVtecsSGo+A9MJnWLnhLp7Xt8h/h0J4zVud69HgXQqGoxu2+jZh1hZAd3A/1zZsw5GRb0nmBdeHS913wA+tCc/tvKI4fhfbeXaivXwMA2NcPhzC6HQThDcEo2bGhzGbITp+CfMcvlnpUdgL8LY7AdbdwdGsVAHeRPer5i8Hn0o8mGhoaGhqaNw367VxDbIk+SVNzjOSfFqGB5R4EO6I1YDaDE/aOzXUcTTqNg0+OAgA47DAwmQ7gchpazrfxEKKzjysMBgNEIj5++WUbMlVlcZP69RsMe3ta3QUotjnQ3L6N7M0brNK9v/oG9iGhVmkOkY3gENkIpsJCyA7sg/buHWjv3ob2bokXPe86MDqIYPfglqXMPVEwCkMiwQsiECri48NQN7DtaJUwGhoaGhqaNxlacLARvbl4UvM0xQhp03+5M28ZlL4QunM/AwDs+80ES+xdo/IHHx/F0eTi+ApMhjMcHax3KeqJHNDP3x1cFhMGgwHbt5dNhrlcLnr3HljrjJ5fB5TRiNxdO6A8fdIq3alVG7j07gs7YfU2DCw+H24D3gMGvIfEhFRc3bYPQlU2gtKSYVcS7F1uL4Eypg969Wz51hox09DQ0NDQvK3QgoONmMzFetbKfNNzctLYCmUyQr3+U8sx09nDZqEhQ52FJTfXwMTwAJcbC2fHwAp5mrk5o5O3C7is4nv34ME9XLlywSpP//5D6AksAJNGg6TJE2EqKPPawwusC3HnrlaxE6qisMiIs7cykJJTgMRUJeT5RYCgHiCoh/bhbvBTpyHwnRYIdnV+ncOgoaF5gxk2bAgRG9tRNnDg+8/1rNW6deOozZt33g0ICCx69lyvXp0ajB8/MemfCHxXvq3Vq5d7ZGamc2fMmJP8svXGxraKXL9+6z0/P/+Xcl8aHz/Bz8XF7ZUEXaOpmvLf3b1794iPHz8iWb16/cPnl3z7qLHgQBAEA4AngBySJI3Py//2QKsovWp0J1daHdv3n/3cMkn5uVh1n4Qdyx12vPcr/QI3dnFCbz83K4EgMzPdIjTUrRuCnj271go3pa8Ts0GP7J/XoeDqFUsaw84O3l9PAj/ANu/KmTINDl5Iwl/3y+wffN0cQPiI0K1lHXhKBCWp9SuvgIaGhqaWMGrUGEsU0JSUJM6gQf0a/PHHX1bRvMtT3aT+1KkLNysrU560tBTOhx++V6/0uKioiMnhcM2lr7ZZs+a/9RPXnTt/cdmzZ6e7QiHncDgcs79/oGbOnEVPHB0dzf9Wn3r37ifv3buf/N9q/9/GZsGBIIguAKYBiADAAtAUwA2CINYCOEeS5NbX00WatwmKMgP6QhjT7lqiPzt8ug4MZuVfxb1J2biam2+VZsdytzpuKAIG1A2qtLzBYMDu3VthNBYHcYuN7QSp1Pdlh1HrMWm1eDz2M6s0cfeecOnZu8oyZjOFtFw15AVFuPNEhntP5chRlEXMey82CG0beoHLoW0VaGhKMZlMtEtnmhfC29tXX17AaN26cdS6dZvvl9+FOXhwr8u/07vXz6VL5x02bVovnTt3cWKDBg0LFQo569SpEzX3+03zSrFJcCAIYgiAnwH8guKozuUtJh8CGArgLRccikX8zFwT0pIU8Pajve7UFMpQBPWGEVZpvA5jqhQajqflVRAaAMBsSseAuo3QUOIIZjVqRpmZ6Thx4nfLcZMmLWihAYC5qAgZK5YCAHh1g+Dz9bdgVOHyVKkuwoHzT5GRp0FimsrqHNuOiQAvJ7wTKUV4oASO9pzX3ncamtdFr16dGnTt2jPn9OkTkuzsbG6rVm3kY8aMS//uu8n+JJngEBgYpJkzZ9FjkUhk+vLLMQEJCfcdDQY9w9fXr3DChG+TQ0LCdEDxKjOHwzXn5GRzEhLuO06fPuuRwWBg/vjjCm+5XMbh8fimXr36Zn/66cjs6vqzdu0q9/37f3NnMIDBgz/OWLFicZ3Nm3feVamUrPj4CUEHD574u1QgOXLkkHDz5g1e27f/en/ZskVeyclPeWw2h7p27bLQxcW1aObMeY9Pnjwm2r//N3c224768suJSdHRMRUfriienI4YMTpl37497iqVkt2tW6/snj37yGbMmOyfmprCj4hopJo9e8FTDodDAcD27Vtddu/e7qHRqO0IIlT97bdTkz09vQwAcPbsGacVKxb7KJUKdnR0jPxZ3yK7dm2T7N69w0OlUrIDA4M0kyZNTfbxqWOT6s79+3cFK1Ys9lUqFeymTZsrJ0+ekczj8SiFQsGaOnWi/6NHiQKTycQgiFD1xIlTkqVSbwNQrHJSr154wd9/33RKTU3mBwUR6pkz5z6VSFyMAPDbb7vEmzb9LNXpdKyePftklW9z2bJFXhkZ6dw5cxY+HT16BAEAHTtGRwLA3LmLExs3bqqxpe+l17lUFSs+foIfj8c3Z2dncR48uO8olXoXTp8++6m/f0AFNa3KUKsL7EaPHl63srJXr14WLFu2yDcrK4Pr4eFVNHbs+JQmTZppSq9FWFiDgtu3bzqlpCTz69VrUDB9+qync+fO8r1+/YrQ01OqmzVr7mNf3+Jo0K1bN4769NORqfv3/+ZeWKhlxsTEyb766ts0JpOJJ08ec2fP/s4vOfkpn8ViUQ0aNCyYP3/pE1v6UNX9uHfvriA4OETdoEHDQgAQicSmfv0GyErHffLkcecNG9ZKs7OzuPb29qb27TvljR07PgMo2xH63/++TNqyZYNUp9MxP/xwaHpYWH3NvHmz/PLy8jjR0e/IJ0/+LgUAdu/eITl8+KBrYGBd7dmzp8VCocgwduyXKW3aRFdQTdi9e4fkyJFDLj//vJUsvS6jRo1J+e233e75+fl2bdu2k0+e/F0Kg8GAyWTCggU/eJ85c1LC4/HNffq8m7V27Srf6naq3nRs7XU8gPkkSX5LEAQL1oLDPQBfvfKevWGUn57evZ5OCw41xKxVQbP1f1ZpTLcAsP2irNJOZ8hwNlMBQ7nI3vnqbWAyHcFkCGAwpmJJ9DSwWewq23pWYGCz2Rgw4EMwa0k8AIqikLtzGwAGRB06gS0Wv5J6NXduI33FUsBUbKfjEBkFr8/HAACK9CZsPUGCzWLi3N+ZCKkjxJOMfOj0ZTY9biI+6rg7IjxQAjsWE+GBEtptKs0L8/DmOUnS/SuvdbXUL6xpXlBkW9nzcxZz/vw50dKlqxONRiNj6NAPwsaNG20/ceLkpKAgQjd27MigrVs3uo0ZMy6zWbMWqunTZydxOBxq4cK53jNmTAnYtu3X+2X1nBXPmjX/YePGTR/p9XpG377dGkyZMuNJ8+Yt1QqFgpWamlytN4YzZ0467dv3q8eiRctJX18//YwZU+qUnouMjNI6ODgahhKiOwAAIABJREFUz5497RQTE5cPAMeOHZbExsZZxnn9+jXh9OkzH82cOffplCkT/b76amxwhw6dcw8cOHZ7z56dksWL59eJjo65U1X7165ddtqwYdv9jIw0zsiRn4QlJNx1mDZt1hORSGwaNmxIyIEDe8X9+g2QnT9/znHTpnXSefOWPCSI0ML582d7T578TcD69VtImSzP7vvvpwSOG/d1UocOnZVbtmxwPX78iGv79h1lJX0W7tjxi+ecOQsfBQTU1f3440rPqVO/DdiwYdsDW+7VmTMnJYsWrUgUCATm8ePHBK1evdxz3LgJGRRlRufO3fLatYt9YjKZMH36JL/582f7Llmy6nFp2bNnT4vnz1/6UCr11o8dOyJ4w4Z17l99NTGdJBN4K1curTNz5ryHjRo11ixePE8ql8srXRFZseJHctCgfg2OHTt781VMAM+fPyf+4YcFieHhEdr4+K/9V61aJp0/f8mTlymrUMhZ8fETgkaOHJ3So0cf+e+/HxDHx08I2rFj7x2xWGICgD///EO8cOHyh2KxxDhs2JCQYcM+DP3f/75KnjVr3tMpUyb6rV27ymvmzHlJpW1duPCn8Oeft97XaNSs//3vs+Bdu7brBg58P2/NmuVejRo1Vq1du5HU6/WMv/++KQAAW/pQ1f2oXz9c88svm6TLli30at68laphw0gtl8u1TA7s7fnm+PjpT0NCwgoTEu7zJ0wYGxwcTGg7depqcZd4//5dwa5dB+5cvnzRcdq0+LoREZGqZcvWJBoMBsbQoR+EXbp0Xt6iRWs1ADx+/FDQpk20/PDh038fPXpIOGPG5MBdu/bfEYnEzzVu/euvi87r129NUKvzWcOGfRj6xx+nlO+80z5/x46trtevX3Vev37rfXt7gXnixPEVDTJrGbbOpOoAOFHFOR0Ap1fTnTeYcpKD2UzbO9QUzS9fWD47DPsZjsM3QtBrqiXNTFFYcicZJ9PlVkKDRnsU/4sYjGXR47Gw9adYGTOzWqHh0SPSSmho3rwN3nvv41ojNJh1hchYvgTKkyegPHkcT78ej8y1q6F9kPDCdVJGI1Lnzkb60kUWoQENovCdJgyfzDmNkQv/wKhFZ3HhThb+uJUBM0XhfpICYiceAKBP2wCsHNcWc0a0wKhe9dGqgSeahbnTQgPNW0evXn1z3NzcjV5eUkNoaJg6OJjQNGjQsJDH41GtWrVRPn780B4A+vcfJHN0dDRzuVzqs8/GZqSkJPNVKpVFH6lx42bKpk2ba5hMJng8HsVisagnTx7x8vPzmSKRyBQeHqGtrh+nTp0Qx8bG5YWEhOns7e3Nw4d/ZqUjHxvbQXbs2BEJUDwxu337llO3bj0tOtchIaEF7drF5tvZ2SEmpr2ioCDfbsSIz7PYbDbVrVtPeV5eLkepVFapP/X++x9mOTk5mUNCwnTe3j6FjRo1yffz89c7OzuboqKaqBITSXsAOHbssDg2toMsPDxCy+VyqS++mJD+6FGiICUlifPHH6edpVKfwq5deyjYbDb18cfDcpydhYbSNg4c2Os6YMCgrODgEJ2dnR1GjRqTmZycxE9NTbZp67JHj9453t4+BpFIbBo8+KPMc+fOiAFALJaYunTprrS3tzc7OjqaP/poWOb9+3cdy5dt376jrG7doCI+n0+1bRsjf/LkkT0AnDhxTNSoURNVixat1FwulxozZnwGg8H4R174zZo1VzRq1FhrZ2eHDh06y5OSnvBftuyZMyedPTw8i/r06S+3s7NDz5595F5eUl15dZ/27Tvm+fsHFJXeWw8Pz6K2bdsVlH53njx5bOWjfNCgD7JEIrHJ29tX36tXv+wzZ06IAYDFsqOys7O4mZkZbB6PRzVr1kJdgz5Uej+aN2+pnjJlxqNHjx7ax8d/HdSlS0zEnDnfe5tK3mMtW7YpCAurX8hkMlGvXv3C1q3byW/evG51r4cN+yyTx+NR0dEx+Vwu1xwbGyd3dXWz/MYfPEiwjM/Jycnw8cfDcthsNtW9e2+Fp6dX0ZkzJ23y6DF48EdZQqHQ5O3tq69Xr0FBYuIDewA4e/aMqFevvtlSqbdBJBKZ3n//w0xb7+ubiq1v/lQAkQBOV3KuMYBHr6xHtQA6lIPtUBQF9U8fW44dh2+sNN/xNBlydMU71AZjKsxmOXRFtyBg2yFIVCygs5hV6wknJNzF1asXLcctW0ajbl3iFYzgn0N1/k9kb1wPAGB7eMBr1BjkbN+KgiuXUXDlMgQRkXDp3Rdcac3c1abOnwPd4+Kf6Dqf7sjjioAS0wS2HRN6gxkhvkLwuXZoXs8DTAYDjYJdaG9TNK+VoMi2sprsBvwTSCQulokth8M1i0RiiwMQLpdnLiwsZJlMJixePF964cI5UUFBPpvBYFIAIJfL7JydnU0A4ObmZihf73ffzX68ceM6z40b13n7+voVjhw5Oq06tRa5XMYmiFDLeanU20p9p3v3XrIhQwbU12g0zMOHD4oJIlTt4eFpaVMoFFn129HR0Viq1sTn880AoNGomUKhsNKVVBcXV0t5DodrFoslVtdFoZCxS/rJCQ4mLKu7Dg4OZoHAwZSVlcXOy8thu7i4WPrNYDAgkUgsx7m5OZy1a1f5rFv3o+WBRlEUIysri22LupK7u6clj1QqLVIqFRwA0Gq1zHnzZvncvHndSaNR2wGATqdjlrc1KX+feTyeWafTMQFAJstlu7q6WuoVCARmBwfHf8QJjEhUdo35fJ5Zp9PZbBhTVdnc3FyOq6ublbqTq6u7Pjc3x7L6JhZLrO61UFgm3HG5ZdemFE9PL335z3K5nA0AX3zxVdrKlUulI0Z8FCoQOJj69u2f1b//IJktfajqfgBATExcfkxMXL7ZbMaFC386zpw5NdDXt45u0KAhedevXxWsWbNCmpaWwjcajQyDwchs0aKlonxbbm7u5b67HHOpSlrpeLVareU6i0QSQ/n3nqurW1Fubq5Ngqyrq1u568Y1a7WFTABQKORsd/ey32b561dbsVVwWA9gGkEQ2QD2laQxCIKIBfA1gBmvo3NvEkXGst8wHQTONiizGYXHlliOBYMWWZ03mM34/sYTGKnyaknbQVFqy/HUZt8+t53du7egsLDMSDcmphO8vWuPLUPe/r2QH9xvOZb07A1J954AAJ+vvoFRpUTuzu3Q3LmN5GmTwXJ0hEu//nBq3tISlflZKLMZGauWI/FWsV1dLscZm727wMBko467I0LriNAu0gtuIjrgHQ1NTdi//1fx5csXhUuWrEr09a2jV6lUrG7d2kdYvxesV6kjI6O0kZFRjw0GA2PLlg2uM2ZMDjxw4PjtqtoQi8WG3Nxsy4QlPT3NavLi5SU1BAUR6mPHfheePHlc0qNHr5xXNsAaIBZL9FlZmRa1K41Gw9Ro1CwPDw+DROJqyMu7YOk3RVGQyWSWY4nERf/eex9kvqh3muzsTEtdGRnpHKFQpAeADRvWuqenp/LWrt2Y4O7uYbxz5zZ/1KhPwmx5b4vFLoaUlGTLSr9Wq2Wq1QWVzpNqw8KKq6ur/sKFc1Z61Xl5OZymTZurqirzPDIzMzil9jxZWRkcsVhsAAB3dw/jjBk/JAPA5cuXHL75Znxw48bN1K+qD0wmE23aRBfUqxee/+TJYz4AzJw5zb979145gwb9+JDH41GzZ8/wyc9XvfBWuEIhY1MUZbm3eXm5nFat2iifU6xahEKRIScnyyIkZWZm1HpjQFv1N+YC2AJgE4DSH/lFAMcA7CRJctlr6NsbRaGhTD2GVlV6PuotY6Fe9wlMqcXvxoPvzMLkBBkmXX1o+Zt2/bGV0KDWHgRFqTE64lOsjJmHlTHz4MARVNUErl69iM2b11qEBoKohw8+GFZrhAazXo+Hnw23EhoCFi2zCA2l2DkL4Tl8FPx/mA+nNm1hKihA9ob1eDhiKLJ+XoeitFSr/BRF4fGE8dCUCA0PBHWwvW4vLB4fg58nxmDax03QP6YuLTTQ0LwAGo2WxWazKZFIbNRqtcxlyxZKq8uv1+sZe/fuEatUKhabzaYEAoG5dJeiKmJi4uSnTp2QJCY+4Gm1Wua6dWs8n83ToUNn2a5d2z1SU5P5HTt2fanJzYsSF9dJfvr0CcmdO7f5RUVFjKVLF0gDA4M0vr5++nfeiVWlp6fyjxw5JDQajdi4cb2bSqW0vEh79OiTu2PHVs8HD+7zAEClUrEOHdpvs/HgwYP7XNPT09gKhZy1desmzzZtohUAoNVqWRwO1+zsLDQpFHLW+vVrvGwfT0fFjRtXnS9fvuSg1+sZy5cv9qIoqlIJQSJxNTIYDCQlPa3WXsVsNjF0Op3lT6/X/2MSR7t2sarMzEzuvn2/io1GIw4e3CtKS0vjxcS0f2HBYceOrR4KhYKVlpbK3r//N7d27WLlAHDo0H5RenoaGwCcnZ2NDAbAZDKpl+nDsWOHhQcO7BUpFAoWRVG4ceOafULCXcd69RpoAECn07GcnJxNPB6PunHjmv2ff/7xUgaB+fn57I0b17sZDAZGyXj47drFvvC1AoDo6HcU+/b96p6Rkc5WKpWsbds2e7xMfW8CNklmJElSAD4nCGIxgBgALigWIE6TJJn4Gvv35sAo/UeBogWHKnlWNQkAFkXFo0hZuUqvhMvGiFBvLLq+BCpT8aJZqDi4yvq1Wi327KnowKtv30EQCBxeouf/HPmXL0F+8AD0WcWqjlwfX3hPmAiWffUTeZaDAzw+/ATuH3wEzd+3kLNzG/Ivnkf+xfPIFbjhnGMYitg8DEo9CgDQMTlY4j8AcU18sTy2bq1YIaOhedPp3buf7Nq1y059+nRpKBA4GIcM+STj+PEjrtWVOXHiqGT16mW+ZrOZ4ekp1U2aNO1pdfljYuLyExPJnC+++JxgMhnUwIGDM8+dOyPhcDgW3/UdO3ZRrly5pE7Tpi0U9vb2L+zT/rvv4n0BYNq0WSk1Ldu2bbuC1NTkjKlTJwZqtRq74OAQ9fffz3kCABKJizE+fvrjFSuW+C5aNNcvOjpGHhwcYtlO7ty5q7KwUMOcPj0+IC8vl2tvb28KD4/M79atp+LZdlavXu5x9+5tx5Urf7LELWjXLlY+btznwQqFgt2kSTPlyJFjMgFg8OCPsqdOnRjQtWtshFAoMvTt2z/r2rUrNrnwDAkJ040aNTZl1qzp/kVFRayePftkicXiSlVL7O3tzf36DcgcM2Z4iNFoYsyZs+ChwWBkxMdPCCrvQnXv3j0ee/fusUwWg4ND1KXeeF43YrHENHPm3EfLly/yWblyia+7u0fRzJlzH5VX16kpLVq0Vn7yyfthWq2WFRPTPq9//0F5AJCQcE+wZs0Kn8JCLcvJydkwfPhnKaXB7V60D05OzsaNG3/yWrVqqa/RaGQ6OwsNvXu/m1W6SzV69Bcpa9eu8l6zZrlvaGi9glat2ijUavUL+z4ODAzSpKen8rp0iWno5ORsnDJlxuNSA+4XZeDAwbmpqSm8jz9+vx6fzzf16NE7OyHhnmNtdtHMsGX7jiAIAUmSNrsZe4PxA/BUJlPXeNfgl4Xfw46RiwJDK0ikweg9+PnRdGsrrq6OLxwcrWD9cMCkxymvWFxxb251bkSIN4RcNhzZxT8YrcGE/Y/34WJmWQCylTHzqqzbYDBg+/YNVml9+gwCm80Gl1vtok8FXmaMLwJFUVBfvwbl6ZMoTCx7Z9iHhsH7y69rVNfTzHycup4GLoeF+/dT0SD1GsLzH4FDWT+HGVMXo2WU738i0N0/fT//Df7NMTKZDEgkDgDgDyDpReu5efNWkrd34HOjBtNYk5j4gDd06Af1zpy5ZOXCsXfvLvXHjfs6+Z+IoExDU10079rOsy5WXxenT59wWrp0YZ39+49W6dXsTSAt7bFLZGSEX2XnbNUFyyEI4hCAHQAOkyT51n1pnk/Zai2tqlQ5hocXoaMYWBwZX+HcV+F+EHOtvSHNvDwbBYYye4ZxjUZVKHfs2EFkZ1s7IWAyWejde0Ct2WFQ37ppiZtQSp1p34Pt6gImzzbHGRRFYf/5pzhwIckqncVkITkyDnFdRoN9+wry9uyCuHtPiOI6VhmbgYaG5s3nyJHfhTEx7VVarZa5YsUS76ioJsryQsPhwweFDAYDrVu3pYUGGpo3lMLCQsbFi+edoqPfUeXkZLM3bfrZq1mzlv+KauGrwlbB4WsA/QHsAaAmCOIAioWIYyRJ/iMeB/5tysQGCmYTLTgAxS5UtzzMAKnSIkiXgS6JO7C0YVlIj2Eh3tAaTQh2tge73CSWoijsSNxrERr6BfVAK6+m4LCsbYZSU5MrCA18vj369Xu/1qjdPI3/Bobs4jhPbFdXuLw7EA7hDcF4jt9vo8mMG4m52HKMBIvJgN5otsRU4Ngx0aO1P4QOHDQJcQfbruTatouBsF3Max0PDQ3Nq2H16uUee/bsqGC7EBISpl658qeHBw/udV20aI4fk8lEaGi9gokTp1hUiYYNG0KkpaXxvvkm/mltcTVNQ/NfhKIoxsaN67x++OG7ADabY46KaqwaPfqL9H+7Xy+DTapKpRAE4YViAaI/gOYAFCj2srSDJMmq4jw8W0cwio2sJQBkAIaQJPnwmTwdAMwG0ADAcpIkvyp3bjqAzwCU+rW+QJLk5zYOwQ8vqKq0beFMsBg5KDC0hKNLXQwY2rhG5WsTtqpEbHuUibsKdaXnZjWuqFNvpsxYfmsdEhVl3nu/jPoMAc5+FcpnZ2fh2LEDAIDu3fvByckZRqMBXC6vBiOpmtel9qFLTkLBlb9gLiyE6txZS7rX6P/BIeL56m0GoxnXyBz8dPB+hXPREV7o/07dGsVP+C+o8AD/jXHSqko0NDQ0NP8Er0JVCQBAkmQGgCUAlhAE4YtiAWIcgA9rUNcaACtJktxKEMRgAD+i2OC6PE8AfAqgH4DKZoqbywsT/zT/deNoM0XhaGqeRWiIST8JmVs9/M32BJvBwPSowApCQ16hDNMuzbVKG93w0wpCw/Xrl3Hv3t+W40aNmkIkKnaU8KYbE8kO7ods/16rNAaHg7or1lSpNlRYZERSVgGK9Cb8eTsDNx9az6kWj2kNJ3t2rdlhoaGhoaGhoXl7eSF/twRB1AUwoOTPE8UB4mwp5wagEYC4kqTtAFYQBOFKkmRuaT6SJB+V5O/1Iv17vVD/ORsHM0Xhcb4WIi4bBQYTfnqQZjkXJr+Ld9p2B0vsjQFVljdbCQ1fRY2Gv3NFl6kFBflWQoNU6oOwsPCX6nvurh1QHD8KYfsOcOn7LpjsqqNOvwwmtdoiNLj07Q9BwwiYVErwiZAKQkNCsgKX7mUhJasAKTkVd2zaRXihnr8EUUS1jlpoaGhoaGhoaP5RbBYcCIKogzJhIQJADoDdAEaRJHnBxmp8AKSTJGkCAJIkTQRBZJSk51Zb0pqBJepMWQCmkSR5qQZlX5ByxtGmF/Z8V+v49Wk2ruflV3quVeafaEtlgiXuXW0dY85MtHyuymuSSqXE/v27AABNm7ZCSEi9F+xxMQXXryJz9UrLsfLkcShPHoeoYye49O3/UnU/S3njZ6+x4+AQ3hAAYPbwhMFkhh2Dwq9nHyMjV4Pk7AIo1dbe/SKDXNAo2BVCBy5C/URg0rsLNDQ0NDQ0NG8gNgkOBEFcARCF4tgNvwH4CsBZkiT/jRn0GgCzSJI0EAQRB2A/QRChJEnKbK2gRFe3ZpSfyzEYcHV1rHkdtQgDzw7T/kyo9ByhSECfpN8AANKh88Gt4lqYKTPe3z3Gcrz93RVgMStXNzp79hiAYnWkNm2aV5rHFrQpqbg55gurtPD5c5Bx4CDy/rwAxbGjMDx9DHNcLNzbx76U56H8ByTufDPJcmyqG4b4Pwpg/9dfyNfoUVhUtd+A8YMaIdRPDA9J1QHuXhVv+3e1lP/COP8LY6ShoaGheXOxdcfhHoCpAE6U7ha8IKkApARBsEp2G1gAvGCjqhMAkCSZVe7zCYIgUgHUB3C26lLWvIhxdBkUClQ6XLnwFP7BLi9Yx5vNeXkBDj+2XGY0dnFCbx8RTLnJKDw4y5IuGDAH+SxX4BmDTYqisOrvn3FfXuYOeU7rqZDLKgaBoygK27dvgNFohL29AP36vf9CBqCae3eRvniBVZpTi1Zw/3godEwmxB8Og2jIp1D9cRo5v2yBOvEhMk6cgfvHQ8Fxdatxe9mHDkK171fL8WZpZ2RQLoBWjwJt2Y6Cs4CDAq0BZorCzE+bwc6OCRcnHphMBmA2v3Zj1/+C0TDw3xjnG2IcTUNDQ0PzH8bWyNEfPz+XTfXkEARxC8B7ALaW/L9Z3r7heRAEISVJMr3kcwSKPSW99iiMzyqP3Lqc+tYJDsoiA+bdTrJK+yzMB94CHgrWfmSV7vDJWjDsrN2nAoDBZMAXZ63jOMxrMx0CdsWoyIWFWuzeXRYFOjq6fY37TJnNeDj8E+u+NYqCx7CRFewZGAwGhO/EQlA/HIXnTiHnzB9I+vZrcH184f7xUPB869jUpjY9wyI0HHFtjgcuoSB8hch6IsesYc3gLq4+AjQNDQ3Nf5Fhw4YQsbEdZQMHvv9cz1rVBRvr1atTg/HjJyb9E4Hvyre1evVyj8zMdO6MGXOSX7be2NhWkevXb71XGl2Zhqa2UKXgQBBEFwDnSZLML/lcLSRJHraxzZEANhEEMRXF7lyHlLR3GMBUkiSvEQTRGsVxIpwAMAiCGAhgKEmSxwDMJggiCoAJgB7AB+V3IV4f1qJDVnrlev+1Fa3RZCU0RLk4oa+/O8yF+ShYO9Iqr8OwDZV6+aEoykpoGBDcG229W1TaHkVR2Ldvp+W4f/8PwLMxGBoAmPV6PBozCjCVbYDZGoWZ7eoKr5HDwG8Xh+xNP0N7/x5SZkwDAHiO+hwOkVGVqjCdvfwESbt/QwvlXQDAzuDeGD2qE0SONYtaTUNDQ0NT+xg1aoxlrpGSksQZNKhfgz/++Msqmnd54uMn+Lm4uOnHjZuQ8ey5U6cu3LSlzaraebburKxM9sqVS71u3LjmXFSkY4lEYn2bNtGKoUNHZgkEAnPr1o2jOByumcEA2Gy2OTw8In/SpOkpQqHwZbRIaP6DVLfjcAjFsRqulHyuDgqATb4ySZJ8AKBZJeldyn0+D8C7ivIf2tIOjW2o9AbM/TvJKq1XsCcaOtjDmHYPhYfnW9IdPlwJBrdqnfzkgjKNs2XtfqjSngEAtmz5yfJ5yJDhNeozZTbj0WfWZQIWLoWds3ON6mFLJPAePwFFqSnI3bUT2oR7FoNqp249ccUhGE8URsjzdchTavFFwiaURmt6EtgMU77uWaP2aGho/nuYTKY33pU0Te1GoZCzRo78OIQgQtWrV6974Ovrp09LS2Vv3vyzR1LSU269evULAWDdus33AwICi/Lz85kTJ44PXLVqmdekSVNtVhWnoQGqFxz8AWSW+0wDoFhGejugKKqC0PDRj7PgHN4AproU9Gl3LelV7TIAgNaghUynwPxrKwAAM1tOqlJouHz5PEiyLLhZ376DatRn+ZHDyPt1l+W47orVYHC4L2XkzPXxhfeXE0CZzVBfv4asXTuRf2g/QgD4sHi4LAxDM10OAMDO2wfS4SMR7CV94fZoaGjePHr16tSga9eeOadPn5BkZ2dzW7VqIx8zZlz6d99N9ifJBIfAwCDNnDmLHotEItOXX44JSEi472gw6Bm+vn6FEyZ8mxwSEqYDileCORyuOScnm5OQcN9x+vRZjwwGA/PHH1d4y+UyDo/HN/Xq1Tf7009HZlfXn7VrV7nv3/+bO4MBDB78ccaKFYvrbN68865KpWTFx08IOnjwxN+lAsmRI4eEmzdv8Nq+/df7y5Yt8kpOfspjsznUtWuXhS4urkUzZ857fPLkMdH+/b+5s9l21JdfTkyKjo6pdNu8devGUSNGjE7Zt2+Pu0qlZHfr1iu7Z88+shkzJvunpqbwIyIaqWbPXvCUw+FQALB9+1aX3bu3e2g0ajuCCFV/++3UZE9PLwMAnD17xmnFisU+SqWCHR0dI3823uyuXdsku3fv8FCplOzAwCDNpElTk3186tikunP//l3BihWLfZVKBbtp0+bKyZNnJPN4PEqhULCmTp3o/+hRosBkMjEIIlQ9ceKUZKnU2wAUq0vVqxde8PffN51SU5P5QUGEeubMuU8lEhcjAPz22y7xpk0/S3U6Hatnzz5W2gzLli3yyshI586Zs/Dp6NEjCADo2DE6EgDmzl2c2LhxU40tfS+9zqWqWPHxE/x4PL45OzuL8+DBfUep1Ltw+vTZT/39AyqoaVXGxo3r3Xk8vumHHxZaIol7e/sYJk2aVqlQ4OTkZG7RorXy4sU/hbb2l4amlCoFB5Iky+vwUQAySZI0PJuPIAg7FBs4v9W8bQ4y9SYzpt94bDnuqUiBw287AACq23egug14NAbsvIJh32NSpXWYKbOVq1UAYDFYEPEqPouMRiO2bfvZKm3gwA/B4diu5qM8e8ZaaFi9Fkx2RTuLFyFfU/yu2pzKxw1JV9ThZ6J1/j34qDMQI7thyec/bQYdjI2G5hVQlJAn0T+Wv1ZDMU6gOI8b6mKzx73z58+Jli5dnWg0GhlDh34QNm7caPuJEycnBQURurFjRwZt3brRbcyYcZnNmrVQTZ8+O4nD4VALF871njFjSsC2bb/eL6vnrHjWrPkPGzdu+kiv1zP69u3WYMqUGU+aN2+pVigUrNTU5GoffGfOnHTat+9Xj0WLlpO+vn76GTOmWAywIiOjtA4OjsazZ087xcTE5QPAsWOHJbGxcZZxXr9+TTh9+sxHM2fOfTplykS/r74aG9yhQ+fcAweO3d6zZ6dk8eL5daKjY+5U1f61a5edNmzYdj8jI40zcuQnYQkJdx2mTZv1RCQSm4YNGxJy4MBecb9+A2Tnz59z3LRpnXTevCUPCSK0cP782d6TJ38TsH79FlImy7P7/vspgePGfZ3UoUNlN7swAAAgAElEQVRn5ZYtG1yPHz/i2r59R1lJn4U7dvziOWfOwkcBAXV1P/640nPq1G8DNmzY9sCWe3XmzEnJokUrEgUCgXn8+DFBq1cv9xw3bkIGRZnRuXO3vHbtYp+YTCZMnz7Jb/782b5LlqyyvPDOnj0tnj9/6UOp1Fs/duyI4A0b1rl/9dXEdJJM4K1cubTOzJnzHjZq1FizePE8qVwur/Qls2LFj+SgQf0aHDt29mZVqko14fz5c+IffliQGB4eoY2P/9p/1apl0vnzlzyxpeytWzecWrZsrWTauICmUChYFy/+KQwJCbVZ0KGhKcXWb/tTAC1QrLb0LA1L0t/uvVgGA6AABoOq1ZsOJjOFKdcfWaUN2rkKHKUCAMB294Ahu3iRpaDAG17dv62yrpMp1o6s3PgumNRsfIV8Dx7cxZUrFy3H/fq9D3t729yQFj5+hNQfZlqleQwdDsfmLV7JBP7OExkW7/q7Qnr7d2PQqsH7oEwmaO7cRv5fl+DSqw8tNNDQvMX06tU3x83N3QgAoaFhamdnoaFBg4aFANCqVRvlzZvXHQGgf/9Blkn6Z5+NzejWrX2ESqViOTs7mwCgceNmyqZNm2sAgMfjUSwWi3ry5BEvLKy+ViQSmUQiUUUXc+U4deqEODY2Lq90F2P48M8yzp8/Ky49HxvbQXbs2BFJTExcvkIhZ92+fctp4sQpKaXnQ0JCC9q1i80HgJiY9oorVy6JRoz4PIvFYqFbt57ylSuX1FEqlayq9Nvff//DLCcnJ7OTU5jO29unsFGjJvmlRrxRUU1UiYmkPQDZsWOHxbGxHWTh4RFaAPjiiwnpXbvGRqSkJHGuXr3iKJX6FHbt2kMBAB9/PCxn7949HqVtHDiw13XAgEFZwcEhOgAYNWpMZvv2Oz1SU5M5tuw69OjRO8fb28cAAIMHf5S5YsVin3HjJmSIxRJTly7dlaX5PvpoWOaXX44mypdt376jrG7doCIAaNs2Rn7p0nkhAJw4cUzUqFETVYsWrdQAMGbM+IwjRw7V3O3eC9CsWXNFo0aNtQDQoUNn+Zo1yytV164MtbrATiJxqbCw+yzDh38YxmAwKZ2ukOXh4al7990ZSS/RZZr/KLYKDtXNlngAbNpOe5uorfPHCkLD71stQgOTz4NLnSygDpCfCmjINCR/NxV1qlhl/zP9LwDAD62nwIlTuX95rVZjJTQMHvwpbF0VqczFqueo0XCMamxT+aooLDLCbKYwZ+t1JKaprM69HxeMdxpJLUHYGCwWHCIi4RAR+VJt0tDQWMMNdZHVZDfgn6D85IvD4ZpFIrElGAuXyzMXFhayTCYTFi+eL71w4ZyooCCfzWAwKQCQy2V2pYKDm5ub1STuu+9mP964cZ3nxo3rvH19/QpHjhydVp1ai1wuYxNE2WqwVOptNZHu3r2XbMiQAfU1Gg3z8OGDYoIIVXt4eFraFApFVv12dHQ0lqo18fl8MwBoNGpmVYKDi4urpTyHwzWLxRKr66JQyNgl/eQEBxOWSbqDg4NZIHAwZWVlsfPyctguLi6WfjMYDEgkEstxbm4OZ+3aVT7r1v1omSBTFMXIyspi2yI4uLt7WvJIpdIipVLBAQCtVsucN2+Wz82b1500GrUdAOh0OmZ5W5Py95nH45l1Oh0TAGSyXLarq6ulXoFAYHZwcKw6IM8rRCQqu8Z8Ps+s0+lYAMBi2VEAYDAYGHZ2dpZlS6PRZDl2cHA0ymR57GfrfJa1azfdDwgILDIYDIytWze5fv75sJDt23+7y+PxavFyKM0/TXVelcJRHCG6lC4EQYQ8k40HoD+AxNfQtzeKWionWCEvKnuXTW8UCA6LicQfizXSxB3agaP4w3Je2DIculMp0KelQnZgH1x6lkWHTi1Ix5yrxZGSewZ0rlJouHHjCu7evQWgOBo0QYTVaMU+e+N6ACiO9txvwEuv9p+6noZfTlT8qn7UOQStG3gWx1agoaGhqYb9+38VX758UbhkyapEX986epVKxerWrX0EZaXAz7CaiEVGRmkjI6MeGwwGxpYtG1xnzJgceODA8dtVtSEWiw25udkWFZn09DQrdRkvL6khKIhQHzv2u/DkyeOSHj165byyAdYAsViiz8rKtKhdaTQapkajZnl4eBgkEldDXt4FS78pioJMJrMcSyQu+vfe+yCzd+9+8hdpOzs701JXRkY6RygU6QFgw4a17unpqby1azcmuLt7GO/cuc0fNeqTMOpZA4tKx+NiSElJtrj302q1TLW6oNJ50j+1++zh4WlgsVhUamoyt3R3BgBycrK4jRs3zQeAiIhG+ZcuXRCOHj0uw5aFOTabTfXv/17e+vVrfB48uM+PiGhU7Q4YDU15qvuG9QawseSPQnEAuI3P/K1BseH0V6+rg28exQ8f3wDxc/K9WZxMl2FBibvVQYGe4LCYyFhdbMzMEwFc5R/FuygsO7i/OxH2cZ8jYN4iOEQ1hvzgfijP/QGKovD56a8tQgMANHIPr7S9Awf2WIQGHo+PkJB6Nj9oKZMJKXNmwahQgO3uAdd3B77UQ9pspnD0ckoFocHFmYe1E9qhbUMvWmigoaGxCY1Gy2Kz2ZRIJDZqtVrmsmULq/WUoNfrGXv37hGrVCoWm82mBAKBuXSXoipiYuLkp06dkCQmPuBptVrmunVrPJ/N06FDZ9muXds9UlOT+R07dlVWVs/rJi6uk/z06ROSO3du84uKihhLly6QBgYGaXx9/fTvvBOrSk9P5R85ckhoNBqxceN6N5VKaVkV79GjT+6OHVs9Hzy4zwMAlUrFOnRov8jWtg8e3Oeanp7GVijkrK1bN3m2aROtAACtVsvicLhmZ2ehSaGQs9avX2OzDWZcXEfFjRtXnS9fvuSg1+sZy5cv9qIoqtKXg0TiamQwGEhKelqtvYrZbGLodDrLn16vr9HLhsVioVmzForVq5dL5XIZy2AwMPbv/02clpbGi45+RwUAH300NLuwUMuKj5/gn5qazAGAjIx09pw533vfu3e3gp9zk8mEX3/dKeFwOOY6dfz+cxojNC9HdapKswEsQPFiez6AGABXn8mjr8xg+r+A0Wj+t7tgM5OuPrQ6dvplPRL/LnMhLfAoO+c4dB0Ero7QlkSo9Rw2EulFy5C9ZSPWJe8FpMXPSKmDJ75t8kWFCT1FUTh+/BCUyuJFpLZtY+HnF2hzX/P/uoisdWstx27v1czrUnl0eiOmrr+CPJVlkQafdgtFy/qe/4lIwzQ0NK+e3r37ya5du+zUp0+XhgKBg3HIkE8yjh8/4lpdmRMnjkpWr17mazabGZ6eUt2kSdOeVpc/JiYuPzGRzPnii88JJpNBDRw4OPPcuTMSDodjefF07NhFuXLlkjpNm7ZQ2Nvbv/AL6bvv4n0BYNq0WSnPy/ssbdu2K0hNTc6YOnVioFarsQsODlF///2cJwAgkbgY4+OnP16xYonvokVz/aKjY+TBwSHq0rKdO3dVFhZqmNOnxwfk5eVy7e3tTeHhkfnduvVUPNvO6tXLPe7eve24cuVPlpdZu3ax8nHjPg9WKBTsJk2aKUeOHJMJAIMHf5Q9derEgK5dYyOEQpGhb9/+WdeuXbHJe1BISJju/+ydd3QU1dvHvzOzs32TLemNEEg2IQRpUgRBQy9CKIoCYkUsgAZE0dBEQIpUQQRBmj8BUZAiSBdeUBGQTlhISO9ld5PtuzPz/rHJJoEQEhREnc85OSczc+/MnbbzPPdpb7wxPnP27BmN7XY7NXDg4Hy1Wl2r25RUKmWHDh2WN27ca9EuF0PMnfvpDafTRSQlTYqsXqNhx47vAqrHdkRFRZu++urrBhWt/eCD6ZmLF88LGTXq2ViHw04GB4faZs+ef6MyHkelUjMrV351bcWKpcFjxrwc43DYSZVK7ejS5cnS6pmZXn11VDOCcFtLAgODbdOmfZyqUqn5Og48DYKoj/nuX0Q4gLSSEhNYtmHn/e2SuVBTBDKsQXCwYfAP9sLg5x9+v/c5527C5Kr6XRi2cQkk1irXWokGUDYB5K9+5UlpeqtQfTH7HIzLVsDH4MLhdgq8/MI8SGupBA0Au3Ztg6EiZqJ//8FQq+ufNIUxmZD6zljPctiUGRCHh9e7PwAYTXak5ZXj8x8uwcXUvMdjB8ehdZT7+/5fURz48/z38HeeI0kS0GjkgNvCnH6v+zl37nx6SEiTu1YN5qnJ9evXxK+88nzs0aO/1igCNmhQ3+aJie9lPIgKyjw8PP8dsrNTfVq1ahle27YG5RDTarUhAKLgjm2oQQMqR/8jCRaGIFYWB47LQooFcDkffiXdxXIepaHL4R8QfjMZJOuemCKEAmiiXKClgLDdM3esg3C5OBmrrm+G5AklXttejJ6/laPE8BkkEybd1qewMN+jNAwc+Ay8veufIpqxWDxKgzK+O/yGj2zQueYUmTB1bW1Jv4Blbz8OueSucWM8PDw8Dw379v2ojI/vbrRYLOTy5UtC2rR51FBdadi7d7eSIAh07tyFVxp4eHgeGPVSHLRarQLAtwB6Vqyq9E+pPqX7r07HKibdboJSyi2AupwPv6vSJ+fdKaCFNisiUq541gc8ChCEO1EE6d8UopZ9b+trsBuRdHK2Zzm2UWs0ntcTae+/C+u1ZNx47WVItNEIHp8IUuR2Xzpzxp1l6amnhjZIaQCA8jNVQn9DlYY9v6Rj+/Ga6a5H9IjCk62C+dgFHh6eh5KVKz8L+O67LbfFLkRHNzOtWPHljd27d/guWjQ3nCRJxMTElldPtzp69Chtdna2+P33k9Lqm6WOh4eH56+gvhaHTwCEAXgcwAm4A6f1AEbCHfvw3H0Z3UOFW0eqFENdroff4mBl3MrN0//7DAAQMvE9MP8337OdCtRCWkudBoZlaigNEd6N8FKsO9Yg8os1yJw7G/b0NFh115Dy1hiUJSTgaqk7qYevrz9UqoYHjhduXO/e/+qv6m5YQWZBORZ9e8FTuA0A+nZohKFP1D+egofn3wrHcWD1NjB6GzgnA8ZgA2txQtI2CJRX/Ysu8tw/3nhjXP4bb4zLv9P2zz9fc+NO2778cmODfOR5eHh4/irqqzj0BTAFwKmK5VydTncawHGtVrsQwCS407L+i6k5c/2wWxyYihiOoIIseGu1CEl8F6yxAJXRDdKhs0Gqak82sfasu4K0t9ALszp9CJKomtEiBAI0mjIdrM2Km++9i9QgfxSUVmUC7NDh8QaNk2NZpH0wCQBA+/re0WUKAFwMi8XfXoC+3I780prZ4yaPaI2o0IZZOXh4/qlwDAumyAJXkQVsuR0cy4GzOME5GDBldsDF1rQHC0hQGgnw34pp4+Hh4eH5i6mv4uAPIEun0zFardYMoPqU8l4A3//lI3vI4G61ODzkMQ5nit2FzYLSroPLv4zy1S96tkl6vg1KXTODIMuxeO//ZsDqqspANPOxyTWUhuqQYgl8Zs7CL7u2AQC0l5OhKtVD1nNAvcfoLClG2vtVmXzDP/6k1naFBit2/t9N/HqloMb6MD85przQFgKKN9Xz/HthbS4wJVYYssphK7WAMdjgzC5zKwcAQJMghBRIkQCEiIKwsQoETYJUCCHwk4GgKRBSGgTvtsfDw8PD8yepr+KQBaAyPc4NAP0B7K9Ybg/AVlunfxdVH90m0b5IvVYEluUeWh/6PZlFAIBGqckQ3JIZmwqLq7HMcizGHZ1cY127gNYQkHU/HrsqlIaWj7SBqqAU1lI90qcnwWfgIKj69Ltr7YX0qR96/o9YtAyEoObxWJbDmh+v4rdbFIak59vAx1sMbznvcsHzz4RzMuBcLEASAMuBs7ncFgICYC1OOHQlcBWa3W0qsoNVWgsJiQB0mDfoMG8I/KQgxIIHVoyKh4eHh+e/TX0Vh4MAugPYAWAxgA1arbYNADuALgAW3p/hPUxUmfi9lO6kUk4HA5G4QYmpHgi5ZlulrAGFyQh5M0AQ0Q6iDs+ClN8ef1BdaXgpdjhCfH3hT9RdMyczsyoNeYtH2gCPtIGjqBDpH76P4u3foXj7dwh9PwmSyMjax7hyOTiHA6RMhqZLV9TYxnEc1uy5WsPCoJQLMejxCDz+SL1r+fDw/C1wnFsRYAw2gOHAGGxw5ZnA2VxgLU63ouC4i8WSJiEMVwIkAVJGg1JL4NvEB3qTDYSAt7Dx8PDw8Pw91FfqfR+AFAB0Ot0mrVZrAjAUgATAWACr7s/wHj4IALTQnUDKYXc9lIrD8qtZAIDHD/8ARQggbJMAUZuEWttWd036pPNUeAkVd80Xn5WVgZ9/PggA6Nt3kGe90NcPTZZ9jvy1q2E+fw5Z89wB1pLoGASNeROkXA5bWhqy5sz09AmbMr3Gvo+dz8GGn6ri/ppHqJH49CP8jCrPQwlrdoAx2sEUmuHMN4Ets4Ozum5rR8qFIBVC0BoJQBAgRJTbUgB4liuNmgRNgfKRgrzlt4VWiEDYaq1FxcPDw8PD80Col9Sr0+ksACzVlnfAbX34z+CJcSAApdpd/Mxhv11AeJAY9VbYrE74B3l51h3MKfH83yTlCui2wXdUGjiOw7vHpwEAxrd8DV5CxV2PabGYcfSo20stICAIPj41C6ZSEgmCx74NxmxG9qfzYM/KhPVaMlITx922L/8XXoLQ18+zXGK01VAa5rzWAf4qCa808DwQOM4dYMxaXG4hnkA1y4DbpYg1O8CW28HobWCKLDUsB6SXEJSPFJSXCIRIAMpb5FYQaAqUWvK3nBMPz8PE6NGjtN269Sp59tkRdy0C2Llz2zYbN269HBHRxH7rtoSE3nETJkxOfxCF76ofa+XKzwLy8nJEM2fOzbjfx/27qH6OmZnpwuHDh8b9/PNvNQoP/hXcz30DwLBhCbFvvz0x87HHHv/H1zmp6134O3j4pssfWqqEV6Go0uLw4AKky402fL/hDzz9chvI5CJcOpuDEwdTPNtFYgGeffVRHM0tBQA8tWUNLAIFsmVdEHQpH9Fxnor3sLlsYDkOSSdnedZFqe6exrS8vAw7drgzLgUHh6Jbtz53bEvJZGg0fSYYkwnFO76D4djPsNDekDjLII3SIvCVV0FrqqpK55WYkfSlO2nXxGEtEdu44SldeXjqC8dxbuuAgwE4wFVggv1qsTvW4G4QAKkQQRDqBcpLBEopBhUgByn8V5ey4eH5z1M9fW59BN+kpEnhJ04cVwsEAk4gEHCNGzcxT5z4flZkpPahiAv95Zf/U3zyyceNd+8+cLFyXV0pguu7z/feS4x6/vmXcsaMeetP7au+JCVNCvfx8XMkJk7KrVy3desPV+rqU184jsOmTet89+3b41tQkC8SiyVscHCw7amnBhUOGDBI/1cc45/GHRUHrVZbhJoJ/epEp9P53b0VoNVqowBsAKABUAJglE6nu3FLm54A5gCIA/CZTqd7t9o2CsAyAL0rxjdXp9Otqe84/wqEIvdlexAWh+x0PQ7uSobN4gQAbFz+W63t7DYX1i//FcSTgRCYnDjv2xfwBZAD6HJ0kCtECAlX4ePfPkW+pbBG30+7fHTXWf2MjJs4duwQAEAuV9SqNHAchzMnMqAvsSD1WhEEAhIMw6JJdDukNG1c1ZAFqK+uQiyhAQIwl7vdLx6tqCHoKDSDCXOnVqVuyZjEMOxt6+6V7HQ90m+U4NLZHAhoEr7+CvgGypGaXASzqcolxC9QAbvdhUYRajgcDCgBCY7lkJNhQKOmasS0CITC2x33QlcTHp0OBgzDghZSIAgCHMuBukf/dMbFIj2lBFazA9kZBqh9pKAoEjabE7kZRiiUYojEAtisTmh8ZfDxV4AkgZSrRXA6XbCYHHA6Gah8ZAhppATLcpAp3MHlJEmA4zjYrO5nTCgSwOV0j/teg//LDDaUFJrAshysFgdIioREQsNud6Gk0ASXiwUtcLvnqH1lkMmFsFqccDoYkCQBi9kBgYCCX5ACMrkQJElAJKZBCUiQJAGb1QmrxQkBw4FmORisLjiKTCAoCqTEnV2IkNDgHAw4OwNXvgn2ZHfCAJCEO3XpLYq/y0sIq48cjIACSZOgKALlFidsVicEtPu+CiQ0xD5SEDQJc7kdQpqArcgEV14ZFN5iqHyk4FgOUrkQHAeIJTQYhkWZwQaCAEQiAVwuFhIp7fkduRMcx3neS4Zh3RYRjoOh1ApjqQUgCFjNDghoCmKJAA47g3KjzeNCyXKATC6ESCzwXNfGUT68Be8BwzAMKIpXKv+rJCQMzU9MnJRrsVjIjz6a0mjWrBnhGzZsvtaQfbhcLtyPWfn7wd69uzUymZw5cuSg5kEpDveT2bNnhP7xxxnvt99+N7Nduw7lQqGQO3Pmd/muXdt9alMcKn+n/82FGet6ElegAYpDA/gCwAqdTve1VqsdCXd8RPwtbW4CeBXuOArxLdtGAGgKIBJu5eOcVqs9pNPp0u/DWKtRdSkqP/j2B6A47N5ysc7tL7/TCWd/ycCF37PBiClwJAF5ruW2dpX78cGjEGhyUKYsQLmyAHHn+2L971VVm0mKwBuTut7Wv1Jp8PLyRqcOfWExOyCVCZF2oxi5GQZ0eDICaxedAMNUXSdXRbrIlEqBrRoMw9UQzqvz29Gb+O1ozUrQgSHeyMs23tZWICDRqUdTaJv7w2p2wlBqgcpHCplcBJblsHXNaZhNDjzauRFIisSJgylQaaSwmB2wV5tddjlZ5GUbaz1GYZ7b0nmxNOe2bRdP5+Di6ZrrJTIaVrOz1nNTaaQIbqQELaRgKrODpAiERajBMBxO/1867DYXAkO8IPcSw6i3oLjADIoibrtWaddr7re40AShiILDziD9RgnqC0EAXkoJbFZnjesBuJ9zv0A55F5isAwHDhwoioS3SgKH3YX8nDKUGWzQ+MogoEnkZRkhFAsgFApQlF+3dZgWUnDeLUC4GiKSgEpAQSEgoRAJ4EuRUFAUCJIAg/qldStnWJhZFrRQAIvDhUKrE6SYgsnihNHBoDyv9tosBHHv5Q9IigDL1N5ZKhdCKhOCZd0fGqFIAKmUhtXqRLnRDqvZAYW3GCRJQF9iAUkRIAnC817dC/2eiUNYBG/Nq42EhN5x/foNLDxy5KCmoKBA1KnT46XjxiXmfPTRlMY6XbK8SZNI89y5i1JVKhUzceK4iOTkqwqn00GEhYVbJ036ICM6upkNcM9+CoUitrCwQJicfFUxY8bsFKfTSa5atTyktLREKBZLmISEIQWvvvp6QV3jWb36c/+dO7f7EwQwcuRLucuXL260cePWy0ajgUpKmhS5e/fBC5UKyb59e5QbN64L2rz5+6vLli0KyshIE9O0kDtz5pTSx8fXPmvW/NRDh/ardu7c7k/TAm7ixMnpXbvGl9V23M6d27YZM2Zs5g8/fOdvNBro/v0TCgYOHFwyc+aUxllZmZKWLVsb58z5NE0oFHIAsHnz1z7btm0OMJtNAq02xvTBB9MyAgODnABw7NhRr+XLF4caDHq6a9f40lvfo2+//UazbduWAKPRQDdpEmn+8MNpGaGhjeoVyHP16mXZ8uWLwwwGPd2uXQfDlCkzM8RiMafX66lp0yY3Tkm5LmMYhtBqY0yTJ0/NCA4OcQJud6nY2BblFy6c88rKypBERmpNs2bNS9NofFwAsH37t+oNG74Kttls1MCBg2sIvsuWLQrKzc0RzZ27MG3s2DFaAOjVq2srAJg3b/H1tm3bmW8dZyVSqZTt0aN36SeffBQBAHl5ufSCBXPCkpOvyEUiMZuQMKTgxRdfLaw8Tnp6mtgtpJ5SvvrqG1m9evXRL1w4N/T8+T+8HA4H2axZ8/IlSz5PBYBDh/Z7f/XV6uCioiJhSEiI7d13P8yIjW1uBdzPdf/+CYWHDx/QFBcXCVu2bF328cfz0hiGIZKS3ot0uVxEt26dWgHA119vu7xt2xbfynO89RyMRiP16aefhPzxxxlvgiDQrVvP4vHjJ+RWPodms5n87bdfVOPHT0xfuHBu4/Pn/5C2bNnaArgV6E8/nRty9OhBjUQiZQcPfrrGtd22bbPm2283B+j1pUKFwsv19NPP5g0fPqoYqLKM9O37VOHOnd/7i8Vi9oUXXs0ZNGho6ebNm3xOnDiuBoA9e37wj42NK1+27IuUShezyMgo2/DhQ+K2b//xgkqlZgDg0qULkkmT3o7avfvgRZqmuTs9hykpN0T79+/1++yzVcmV5wEA7dt3NLVv39FUuTx69Chts2bNTRcvXlCkp9+Url379RWGcRELF84LTUtLlSkUXq4XXngl56mnEvQAYLfbiSVLPg0+efK4yuVyku3bd9S/996ULIlEwgHAl1+u9P/hh+/9CQIYNeplj4Bx7txZaV3v/p2evb+aOyoOOp1uxl99MK1W6wegNYAeFas2A1iu1Wp9dTqdR7rU6XQpFe1rc84fBuBLnU7HAijSarU/AHgawIK/erzVqfy9I0CAKHO7A9XmquRysbh0Ngf+gQr4B3khL9uIwBDvBs00V840MkxNAWHoi61x42ohLvyeDQAYPqYdRGIBHotvgo5PRmDDqZvIAwthmRMdHAeR5f84TkWeg9luhfZClW6mLAmGssRdx4G9RTdkGQ4r5v6Mdl3CERKuwv4dV0HIUwECCAwMhl0fiu/W/wGgKi0tAFw8UyU8B4UpEdcmCCofGYryy5F6rQih4SrEtg5yz7xzHMzldvxxOgdnT2fhJljIQWBIzyi0bR2Mm7pi7N9R08pYm0Bfeb2P7buOY/uu17q9kl+OVCki+pIqxSokXAmFtxitOoTBUGKBqdyO4gITOj4ZAaFIAJblwDIsUnXFKMovh5e3GAEhXijKNyEq1h8uF4OLZ3KQdr24wsLCocxg9ezfx18OACgtMoNlOdhsTlz+IxfV0V2qKT9kpJZ6/heKKNis7ucsKtYPfkFe0PjJQNMUrBYnTGV2SKQ0wpqoPTPxJEmgpNAMh4NB0yhf5OUaIZHSkMpFKJ4M/9kAACAASURBVCk0ITfTCJZlYSqzw2FnYLc5wXJiqDUyiCSCGrP++TllyM8pA0mSoGkSLhfrUTAU3mL4BymgL7agvMyGwFBvUCQJU5kdLR4NRkgjFQQ0BalcCHuFNcPlYuEf7AWapmC3uUBSBMoMNvfsvZCCWErD6WAgpym4csphzTAAFieEtprvmlVIwiajYaYp2EhAJqaRV2KB2WQH4WIhFpAQcQRsTgZ2hwsOIQWBvwwcQaC0yAyxnIaqkTdM5XbIgwTQhishV4g81iGXi4XLyUCmEEGploBlONjtLrAsh3KjDYyLhUwudFtOhBREYhqmMhtKi9xyg83qrLC2OEEQ7mtFUe7rV3mfDKVW2Czu+wXCbcHUl1ggEgvgF6iAl1JccY9cCI/UQCIRwmCwQKWReuKaxBL39aq0fqp9ZRCKBBVKGQeLya0gEyQBkVgAb9XDE2tx7doVTWrqdZ+7t7x3mjSJKo6Ojq23Jn3ixHHV0qUrr7tcLuKVV55vlpg4Vjp58pT0yEitbfz41yO//nq937hxiXnt23c0zpgxJ10oFHILF84LmTlzasQ331R9uE+cOKaePXvBjbZt26U4HA5iyJD+cVOnzrzZocNjJr1eT2VlZdSZS/ro0UNeP/zwfcCiRZ/pwsLCHTNnTm1Uua1VqzYWuVzhOnbsiFd8fI8yANi/f6+mW7cenvM8e/aMcsaMWSmzZs1Lmzp1cvi7746P6tmzT9GuXfsvfvfdVs3ixQsade0af+lOxz9z5pTXunXfXM3NzRa+/vrLzZKTL8unT599U6VSM6NHj4retWuHeujQYSUnThxXbNiwJnj+/CU3tNoY64IFc0KmTHk/Yu3aTbqSkmLBxx9PbZKY+F56z559DJs2rfM9cGCfb/fuvUoqxqzcsuV/gXPnLkyJiGhqW7VqReC0aR9ErFv3Tb1m448ePaRZtGj5dZlMxk6YMC5y5crPAhMTJ+VyHIs+ffoXP/FEt5sMw2DGjA/DFyyYE1YpZAPAsWNH1AsWLL0RHBziGD9+TNS6dWv83313co5OlyxesWJpo1mz5t9o3bqtefHi+cGlpaXC2o6/fPkq3fDhQ+P27z92rj7WAJPJRB48uE8dHt7YwrIsJk16u2mHDp0Mc+cuupmbm0MnJr6lDQ9vbHviiW5l7nv4u/KDD6bdnDNnQZrdbicmTXq7iUQiYTdt2nZFLpezZ86ckgFuIXjhwnnhs2bNS2nZsrV5x47vNElJ7zbdunXnZZFIxAHA8eNH1YsWLb8hFovZ1157Mfr777f6jBjxQtHs2fNv3OqqVBfTp38QrlSqXN9+u/OyxWImJ04cF7l16/8clQL+/v0/KkUiMdOv3wD9oUP7NXv27NRUCtxbtnzte+bMKe81azZdlclk7HvvJdbwj1apNK7585ekhIc3tv/220l5UtJ7kc2bt7C0aNHSAgBGo4E2Gg2CnTv3X/zjj9OypKT3I5s3b2F+7rnniy9fvii/1VWpksDAIGdkpNb00097Vc89N7IYAPbt26Pp2LGTnqZprq7n8NSpX7w0Go2jutJwJ44ePayeP3/xjaZNo2xms5l6/vmnY59//qWcQYNW30hOviKZNOntqKgorVWrjbEtWjQvJC8vV7Rhw+arAgHNJSVNivj886VBEydOzjl69JDXjh3bAhYurHz3pzTo3X8QPGjbVyiAHJ1OxwBARUG53Ir1t09L104YgOqBSZkV/e8rVcZ9DjC447ocdhfMJjtk1eoJ6C7le2bLpXIhLCYHGkf5oPfg2Dvum2U5pF0vRoTWB6VFZnz71VkAQOMo9zdVG+eP+H7RAADfAAU6PBEBi9kBuaLmt+c65VY0el5Yg5C2LPR9pDDcKAVEQHKrA/DNbQqLQo+wlDY1+nUfEAOFtxh7t11C4ygfXLuYj9+Pp+P34+kgaSskCrcwlPKHBOCqLHOVSkN1hr7YGr4BVUHWKo0UUbH+7ivHccgsKIdcQuPSzRJsOF11G+e98zikFVlkIrQ+eGNyV08fACgpNKMovxwxjwRW3QmOq5ipT8P5U9l3vL6j3uqAlGtF+P14Grr1j0FwIyUyU0oR1lQFkZj2tKtNqCJJAiRJQdvcH9rm/p71foFuwY0WUmjfpTHad2l8W987YS63o9xog0IpBscC5WU2pCYXwS9IgaYxfiAI9/lWzkhXXoP6uJhIpO5vXFCFq5dKI4OLrVJAfQMUNe7PvWAxOSAUCyD4k2lBKzOSqdUSeBMEmEIz2DwzqCIz7EXu32mhgATlJYKgiRyCIEVFwLEAyluOfbdMYH8WSkBAKnBf21vfu0okUvpPX9u6aMg5Vl7b6s83z91JSBhS6Ofn7wKAmJhmJm9vpTMu7hErAHTq9Ljh3LmzCgB45pnhng/1m2+Oz+3fv3tLo9FIeXt7MwDQtm17Q7t2HcwAIBaLOYqiuJs3U8TNmjW3qFQqRqVS1SmIHD58UN2tW4/iSivGa6+9mXvixDGPqahbt54l+/fv08TH9yjT60upixfPe02ePDWzcnt0dEx5pQAaH99d//vvv6rGjHkrn6Io9O8/sHTFiiWNDAYDpVQqazX7jRjxQr6Xlxfr5dXMFhISam3d+tGy8PDGDgBo0+ZR4/XrOimAkv3796q7detZUincvfPOpJx+/bq1zMxMF54+/bsiODjU2q/fAD0AvPTS6MIdO77zBNvt2rXDd9iw4flRUdE2AHjjjXF53btvDcjKyhDWx+owYMCgwpCQUCcAjBz5Yt7y5YtDExMn5arVGqZv36cMle1efHF03sSJY7XV+3bv3qukadNIOwB06RJf+uuvJ5QAcPDgflXr1o8aO3bsZAKAceMm5O7bt6debth3YufO7QF79+72o2maa9o00jxlysz0c+fOysrKygRjx76TBwDh4Y0dvXr1LTp0aL+68r41bRpl7t27nwEAjEaD4MKFc967dh04r1KpGADo2LGzCQC2b9/m26tX36I2bR41A8DQocNKtmz5OvCPP07LKtsMGjS0oNIK1K5dB2NKyvUGzyAUFhYIzp//w3vfvqPnJBIJJ5PJ2CFDhhXs2bPTp0px2OfTqVMXPUVR6NatZ+kXX3wW6nQ6s2ma5o4dO6pKSBhS7Z69kJ+U9J7nB7N7956e2cGOHTubWrRoWXb27Bl55bMFuO+HSCTiOnbsbGrduo3xwIG96sjIqLy7jT0+vkfpkSMH1c89N7KY4zicOHFM/cEH028CdT+HBoNB4O2tquE+8NRTPVvYbDbS6XSSmzZtuVz5rHbr1qOk8n09fvyIl6+vn/3pp58rAYC4uEesHTp0Mhw48JMqKio67+DB/T5r1266WmkBGTXq5bxZs6ZHTJw4Oefw4YPq+PiexTExsTYAGD36zdyTJ/+v3u/+g6BeioNWqz2Nu7gt6XS6dn/JiB4AGo28wX0kosrK0QRUXm6f66ybpTh1LA3DXmoLbXP376Gvb5XgYKlwL0m7Xlxj/a3MnLgHANC1ZxTOn87yrE+77lZQeg2I9WRyuhM55VWz3BIZi5AxSzH5sDsdakJMLwxvkQAX44KAct9yjuNw4XQ2Gkf6eATmuJZuK4S53I7lc4/C6TJB4ut+Hq3FoQDnNo29PL4TgsOUOLL3GhiGQ88Bzdz7ZLk7Vqc9cSEH8zaeuW39U49H4LWEuFp61MTPzwsxzQNr3TbgmZYY8EzLOvuHR/ige98Yz3JIqKqO1veX2p6FFq1Caiz7+Xnd1uavPN6f22H9mrEuFpZ0PexFZjAmBwiaAmt3gXWy4Bj3n8toh0NvBdiKnxeSgFAtgVf7EMijfCDyk9XbJ/8vP8+HkH/TOUZHx5Y0xBrwINBofDxCglAoYlUqtcd/TyQSs1arlWIYBosXLwg+efK4qry8jCYIkgOA0tISQaXi4OfnV0PY+OijOanr168JXL9+TUhYWLj19dfHZtfl1lJaWkJrtTGe7cHBITUE6aeeSigZNWpYc7PZTO7du1ut1caYAgICPcdUKlU1xq1QKFyVrg0SiYQFALPZRN5JcfDx8fX0FwpFrFqtqXFd9PoSumKcwqgorUdIl8vlrEwmZ/Lz8+ni4kLax8fHM26CIKDRaDzLRUWFwtWrPw9ds2aV58eP4zgiPz+fro/i4O8f6GkTHBxsNxj0QgCwWCzk/PmzQ8+dO+tlNpsEAGCz2cjqsSbV77NYLGZtNhsJACUlRbSvr69nvzKZjJXLFX/KJ3ngwMH5t86EX758UWow6IU9enTxfLg4jiW02hjPzICPT9U4cnNzaJlM5qpUGqpTWFggPHbsiObHH3d5FByGcRGFhYUeS4lG4+s5X5FIzJaUlDR4RiE7O0vIMAwxYECvR6qPWaNx3+Ps7Cz66tXLitdfH5sNAD169DZ89tmiRocPH/Du3bufQa8vpavfs6CgkBoZgo4cOei1YcNXQfn5uWKW5eBw2Mnw8AiPYCOTyVwymcwzC+bn5+8oLi6u13n06tVXv2rV8rD8/Dw6LS1VRBAk16HDYyag7ufQ29vbZTDoaxxj9+4DF10uF554okMbrprvnZ9fgOfc8vLyhCkpN2TV7y/LMkSXLk+WFBcXCRwOO/naay9WCSTgwLIsAVS++9Gedz80NKxB7/6DoL4Whyu4XXFQAXgMgBXA4XruJwtAsFarpSqsDRSAoIr19SUTQCMApyuWb7VA3JWSEnfQZkOoekA4lOlNEIoEyM1yK8hb153xzJBnZZbW2v/EkRvQVstsVBvHDtTubuNkmDvONLIchylnqrIr9d61CRINkG51j1dOy9AjsFut/YMbK+FwuW7b5uurwFMjmmLfvp0AAKFQiOcTe3sCZgGguNiEFu3c79ndZkFTc42Yt/FsjXUiIYUpo9oi2Ed2X2eK78T9nqF+WLif58kxLNhyB0gvEQiSAOdkwJRawehtYE0OOFL1VVmKaNJdAZkACJEABE2CoAgQYhrCSDUEvlJQGilIb5HbnQ1AOTiUF5vqHMODOM+Hhb/zHEmSuKcJl38jO3d+rz516hflkiWfXw8La+QwGo1U//7dW3I1HPiJGh+YVq3aWFq1apPqdDqJTZvW+c6cOaXJrl13dg9Rq9XOoqICj+CXk5Ndw10mKCjYGRmpNe3f/6Py0KEDmgEDEgpv38v9R63WOPLz8zwmOLPZTJrNJiogIMCp0fg6i4tPesbNcRxKSkqqCbM+jueeez5v0KChtX8070JBQZ5nX7m5OUKlUuUAgHXrVvvn5GSJV69en+zvH+C6dOmi5I03Xm7G1SNQSa32cWZmZnhm4y0WC2kyldcqJ/2ZJAOBgUEOX18/+/ff77l8pzbV9x8UFOw0m82C2qxEvr5+ztjYuLw33xzf4EBkgiDqLQgFBQU7BQIBt2/fkfO1uWbt3v2DhuM4JCW956n66nQ6iZ9++lHTu3c/g1Kpct5yzzzPjd1uJ2bNmt5k4sTJ6T179jHQNM29886bTarfM7PZLDCbzWSl8lBUVCgMD29srTiPOseuUqmYFi1aGvft26PKyEiXPP5419LKPnU9h0qlyrVq1Yqw6rEad6L6GPz9A5wxMbHlX3zx1Y1b27EsC6FQyG7YsPlKUFDwbQK/Wq12FhRUvfvZ2VkP3btfL18DnU73ok6ne+mWvwQAEQBSAfxSz/0UAjgP4LmKVc8BOFc9vqEebAMwWqvVklqt1hdAAoDvGtD/nsmUkjgSGwynywWhiKoRt1AZj/D78fRa++ouV/mxMy7Wo4ikJNd+zwePagUAeLxn0zrHVF1pAICAvExktG6KKb/MAQCMjHm6zv61wbJsNaVBhGeffREEQdTIGFRfLqQUY3aF0vDO0y3w1eR4fDU5HisndEWwj6zB++P5++AYFqzNBcbkgDO3HOU/6FC+UwfjN5dg/PYKjN9chumnVFhP5cB+pQikXAhplzB4PdccyuFxUD7fAt4j4uD9dDN4JURD8ZQW8h4RkHYIgbCJGpRSzGf84XnoMZstFE3TnEqldlksFnLZsoXBdbV3OBzEjh3fqY1GI0XTNCeTydhKK8WdiI/vUXr48EHN9evXxBaLhVyz5ovbzK09e/Yp+fbbzQFZWRmSXr36GWrbz/2mR4/epUeOHNRcunRRYrfbiaVLPw1u0iTSHBYW7njyyW7GnJwsyb59e5Qulwvr16/1MxoNntnbAQMGF23Z8nXgtWtXxYA78HbPnp31NgXv3v2Db05ONq3Xl1Jff70h8PHHu+oBwGKxUEKhiPX2VjJ6fSm1du0XQfU/n176P/447X3q1K9yh8NBfPbZ4iCO42r9UdJofF0EQSA9Pa3OeJXaaNWqjVksljCrVq0IsFqtBMMwSE6+Ij537mytrgUBAYHORx5pZfzkk4/C9Ho95XQ6iV9/PSkHgISEIUU//bTH7+zZ0zKO42A2m8lDhw54l5eX31W+8/HxdZlM5QKj0XjXj3tAQKCzRYuWxvnzZ4eWlZWRLMsiLe2m6NdfT8gB4PDhAz7PPDM8b926/12p/EtKmpF6/vwf3qWlJVTXrk/qd+7c7ue+Z3rqm282eGZSHQ4H4XQ6SbVa7RQIBNyRIwe9Ll48f5vZffnyJUEOh4P47bdf5GfPnvHu2bOPHgCUSrUzPz+3zvvQvXvP0kOHDmh+/fWkqnfv/h4loa7nMDIyyt6jR++imTOnRhw7dtSr8l6dPn2qzlmU+Pgehry8XPH27d+qnU4n4XQ6iXPnzkqvX78mJkkS3bv3Kl60aF5oUVGhAHBblH7++bBXRd/SI0cO+uh0yWKLxUJ++eXK257fv/vd/1MxDjqdzqTVahcCWA6gvilRXwewQavVTgOgBzAKALRa7V4A03Q63RmtVtsZwBYAXgAIrVb7LIBXdDrdfgCbALQHUKnJzdTpdLdF/98PDgQKYRSR0DvcPsRlhqpcLqsX/B9eGNfRszzs1bbYusbtmqPSSJGTYQDLcvh27RlPcG7/YXE4uDMZANDxyQj8WhEbIZML4R/k5bFi3IlsU9Xx43/ahpBM9yXZIK5KltFcE3Nbv7tx9ao7xk+j8UW/foPu0rp2MgvKsWrXFeRVnOszTzZFiyb3NQ6S50/AORm4CsxgrS6wZXawJnehM9bsdKcvdTKA85ZsPiQBYbQPCBJg7QxIKQ3KWwTKRwZCIqi1rgGvGPD80xk0aGjJmTOnvAYP7vuITCZ3jRr1cu6BA/vqdOA7ePAnzcqVy8JYliUCA4NtH344vc5vVnx8j7Lr13WF77zzlpYkCe7ZZ0fmHT9+VCMUCj0vYa9efQ0rVixp1K5dR71UKr3nVFsffZQUBgDTp89usJ90ly5PlGdlZeROmza5icViFkRFRZs+/njuTQDQaHxcSUkzUpcvXxK2aNG88K5d40ujoqI9JsQ+ffoZrFYzOWNGUkRxcZFIKpUyLVq0Kuvff+BtKS5Xrvws4PLli4oVK770zOA+8US30sTEt6L0ej396KPtDa+/Pi4PAEaOfLFg2rTJEf36dWupVKqcQ4Y8k3/mzO/K+pxPdHQz2xtvjM+cPXtGY7vdTg0cODhfrVbX6jYllUrZoUOH5Y0b91q0y8UQc+d+esPpdBFJSZMiDx8+ea6u41AUhU8/XZqyePH8kCFD+rdwuZxEYGCw7ZVXxtyeuq+Cjz+em7ZgwSehI0YMae5yuYjY2Ljyjh07mVq2bG1JTHwvfcmST8Py8/NEQqGQjY6OMbVv3/Gu5snISK2tU6cupc88MzCOZVls3LilztoHM2fOTV+yZEHwiBFDm9tsVtLPz98xbNiIvLNnf5cVFxcLhw9/vrC6m1vPnn2Ma9eusu/Zs1M9fPiooqysTPHLL4+MlUgkzJAhz+RfuXJZAQAKhYJ97bU3Mz/+eFoTp9NFtG37qLFNm3Y1MqJ4eyudCoXCNWBAzxZCoYgdO/btjMp6GIMGDS2eOvX9Jj16dGkZG1uVbao63bv3NixdujBco/FxNG8e53GButtzOHXqzMwNG9b6rVy5LGTGjA/FUqnMFRgYZH///Sk3Q0LCan02FAoF++mny64vXfpp6JdfrgxlWY5o1CjcMm7chCwAmDDh/ezPP18a9NprL8aUl5cJVCq1o1+/AUUAyuLje5TduKErSEwcqyVJghs16uWc6vFNwF/37t8rRH3Md3Wh1WpHwJ1etV4v5t9MOIC0e3FVurRuKw41bYkiMYkXLXk4/Gvd/Ue83g6p14rAMBxcTgbnfqvbG+v197vAZnVi33eX0XNQ7B0DMKuz8moWssw2tD+xHzFX3ErKmgQNzFIKclqGD9q9A6XIu/4nWcHGjasBAMOHv9zg3NEcx2Hbz6n46VTVN2hI1wj06xje4HHcT/4Lri3A7efJsRyYEisYgxWs0Q5XRfVjttzudiWqgBBSICQCUEoxQBIgBCQICe1eLyBAyoSgfKUg71KL4EHxX7ifD4mrUmMA6fe6n3PnzqeHhDS5a9Vgnppcv35N/Morz8cePfprjWJjgwb1bZ6Y+F7Gg6igzMPzd1Fbobr/Ovf73c/OTvVp1apleG3b6hsc3beW1UIAMQASARy959H9Q+BQ5bXKMAzu5uXlpZSgVYcwAEBpsblOxaFpjC8IgoBEKsTgUa3rPaYcs9viUKk0nGgpg1lKYUBEb/QKv7U0Rv1wOKoU6HspODPvm3O4nuW2nD3RMgijekff0zh47g2OdccRgOHgyjOh5HopLMVmsGYHOJs7BqE6pLcIpIQGpZGADvUGKaNBqSXuWATeOsDD87exb9+Pyvj47kaLxUIuX74kpE2bRw3Vf5P37t2tJAgCnTt34ZUGHp7/EH/3u19fyXAPKmTnW9Y7AewEMPavHNTDCQeqIj7c5XQCuLNFIK5NTZdXtY8MIrHAk/9+xOvtUFxg9tQq6DGwWYNH42RZsABk5UZIIoJRGFqEs8HumIGGKg063VWcOfNrhULkpm3bDg0e04HfMz1Kw+QRrREV+k8wQv3zYE0OuEqt7urHDAdGbwWcLFi7C0yRxaM4AIAZAEgCpEIIUkJDGK0BpRBBECAHqRCCoPmKtjw8fwcrV34W8N13W26LXYiObmZaseLLG7t37/BdtGhuOEmSiImJLa+ecnH06FHa7Oxs8fvvJ6X9myvU8vDw1ORhePfr5aqk1Wob1bLaBqBQp9Pdj+rS94tw3KOr0oV1W3CycUtkyygM+mUTzli73bHt4FGtPAWaKnE6GJw7lYU2j4WBokhwHAfdpQKENVFDKqu1vkydbLqRi2SDGY8d+xHB/lewJtgdUza7U1Kt7kkcx8Fms2Lbtq8BAEqlClptLLy9lThwYE+NtiqVCr16DYRQWL9xZeSX46P1pz3Li8Z2glLe4HixB8o/ybWFYznAxcKRZoD9WjFYwy11kkkCpJQGKAKURgKCpkAIKVBeIgS2DUGx3vyvtx78k+7nvcK7KvHw8PDwPAj+tKuSTqdrULrTfyccBBW6hosQYODwFtj5Te3udmQttQxoIYV2j4d7lgmCQHSLutOz3olypwvJBnea3/DUq1jZuioRRW1Kw/Hjh5CefrPGOoNBj1OnTniWQ0PDERfXEj4+fg0SULYdTcG+avEME4Y98tArDQ8rnIuF/Vqxu6aBgIQrqwyuYgvAcR4LAiGlIXrEHwJfWYXFgAQhFtxRMSBp6l+vNPDw8PDw8PA8GBrkxK7VarUAggGIb92m0+n2/lWDehjhAFCVMQ4kBX9pWY3tIeEqZKe7k0Ewrvsb5H40151JTGoqQ6GqynKy/Ml5NdpxHIdNm76ssc7X1x8dO3ZBcvIl3LhxDQAQHByGJ5/s2eBx/HY136M09H+sEQZ3aXKXHjzV4ViuIvbABbbMAXtyEZiSqhgEQiIAHeYFgiJBSAQQ+MshCJDfscgeDw8PDw8PD8/9pL7B0XEANsMdDF2b1MIB+Jc7S3OgKty6nAIFrLr/w3OvDYa+2ILQCDVIkkDa9WL88Usm/IP/uqq/tZGRVwhQIjzzv8+werAPJrR+E02U4be127Jlvef/rl27o1GjCM9yx45d0LFjl3ofk2HdytDSbRdxOa1mnZQFbzwGjfdtuuR/Bq7CIuAqssChKwbnYMAYbOCsLoAkQClFYEptIL1EYMvsIESUu4qy6ZZMbjQJSYdgCBurwJqd7mJovJLAw8PDw8PD85BQX4vDV3AHQvcHkALgruXg/3UQVRaHLKkMZw1n8K56JJTqqnotTaJ90SS6znTefxqW45BHieBlKMGp5lIMYMS1Kg3p6TfhdLqLEj733Eug6QZXmAcA6MvtmLji5B23938s/D+nNHAcB9Zgg+18AZyZxtobEQCpFAMsB87JgtJIwNpcIL3dblyUlwiCEAVIuRCkWABCJoTAT+ZRFKh7KLbHw8PDw8PDw3M/qa/iEANgSEUBtv8kHABXxeQvQQiQJml4QPOfxcW68O6JdZBKnkCjtGt4ROVAdJdxNdrY7XZs3brBs9ynz8B7UhpMVidmrPsdpWX2GuspksCnb3WCl5T+V/jOsw4G1l+zwdldFa5AMhBSGqRM6I414DgQNAWm3A77pUI4btxWlR4AIIr1BektBh3qBVL8cNQ24OHh4eHh4eH5K6mvhPM7gLD7OZCHHgK44VVxuYgHLxgyLIPEY59AIX8WABCRfQahoQxIb/8a7X76aafn/6ZNtfD1rbm9PjhdLMYv/T/PcqtIH4wb0uIeR/73wnHujEQETYG1OGG7UAC2zA6TxQnXLUqRK890h73cjiBYAVG0DwSBchAUnw6Rh4eH526MHj1K261br5Jnnx1x18xanTu3bbNx49bLERFN7LduS0joHTdhwuT0B1H4rvqxVq78LCAvL0c0c+bchy5hzMNYJO2fcu14GkZ9JeDXAGzWarUWuIu9GW5toNPpLH/lwB5qyHtz+/kzfHDyYyjkIzzLTZUmKEavr9HG6XTAaHTfmr59E+Dj49egYxjNDhw/n4NSk9OzbuWErhD9Q91mOI6DceNdfkNJArL4cNDBXmCtvKOGogAAIABJREFUTrjyTWCKrXDlmyAIkLmzG+WZQKklEPhKIfCXg5Q/eGsTDw8PD8/fyxtvjMuv/D8zM104fPjQuJ9//u3snYqlJiVNCj9x4rhaIBBwAoGAa9y4iXnixPezIiO1tlo7PKQkJPSOMxqNNEmSnEgkYlu1amP84IPpmXK5vN6ZYKpfO55/NvVVHIrhzt29sY42/0zpsr5U98p5wBaHYmsJ7KzGc7OeXb8I0uHdb2u3efN6AED79p3rrTS4GBbzvzmHlJzbffVXvfsEaMHDN5vOcRzgZGG7VACBvxwgCFBqMTgXC9bsBFNiAWt2wpFcNalF+cnAlFohaR0AYbQP/Py8bks5S0poCBurgMaqmgds9SDOioeH598IwzCgqH/355HnziQkDM1PTJyUa7VaiVmzpjWaM2dm+Lp1/7v2d4+roXz00Sc3unR5ojw3N4dOTHwr6osvlge+++7knL97XDwPnvpKwF8D6AjgU/xXg6OrU2FxuKFPRaTq9hSkLqcdp/Z9jUe6DIBc+eeCpRmWwfRf50Eq6goAGPq/zyBxWUGHt/a0cTqdNeIatNo7V6K2OxhMXHESYf5yWGwuZBbe7p4jl9AY3iPyoVEaGKMNjN4G+6VCMKXWGtvsl4vu2t97VIt/RTwGDw/P/SMhoXdcv34DC48cOagpKCgQder0eOm4cYk5H300pbFOlyxv0iTSPHfuolSVSsVMnDguIjn5qsLpdBBhYeHWSZM+yIiObmYD3LPMQqGILSwsECYnX1XMmDE7xel0kqtWLQ8pLS0RisUSJiFhSMGrr75eUNd4Vq/+3H/nzu3+BAGMHPlS7vLlixtt3Lj1stFooJKSJkXu3n3wQqVCsm/fHuXGjeuCNm/+/uqyZYuCMjLSxDQt5M6cOaX08fG1z5o1P/XQof2qnTu3+9O0gJs4cXJ6167xZbUdt3Pntm3GjBmb+cMP3/kbjQa6f/+EgoEDB5fMnDmlcVZWpqRly9bGOXM+TRMKhRwAbN78tc+2bZsDzGaTQKuNMX3wwbSMwMAgJwAcO3bUa/nyxaEGg57u2jW+9NZ6s99++41m27YtAUajgW7SJNL84YfTMkJDG9VLvrh69bJs+fLFYQaDnm7XroNhypSZGWKxmNPr9dS0aZMbp6RclzEMQ2i1MabJk6dmBAeHOAG3u1RsbIvyCxfOeWVlZUgiI7WmWbPmpWk0Pi4A2L79W/WGDV8F22w2auDAwTVmyZctWxSUm5sjmjt3YdrYsWO0ANCrV9dWADBv3uLrbdu2M99pvBKJhIuP76mfPXuGJ73hoUMHvNetWx1cUJAvkkqlTPfuvYvHj5+QC1RZNN55Z1L6xo1fBTkcdnLAgMEFb745Ph8AKhWR06dPKZVKlbNHj9413L+uX78mXrBgTqP09HSJWq12vvLKmOyePfsYAfczKhKJ2YKCfNG1a1flYWHh1rlzF6asXbs68OefD2m8vZXOadM+vhkX90jNDy6AoKBgZ5s2jxozMtIkAHDgwD7vtWtXhZSWltLh4eHWSZM+zIiKir7NolL92gHAqVO/yr/44rOQ7OxssUQiZp9//uWcpk0jrXU923d9KHgeCPVVHJ4EMFqn031zPwfzUFNN8HTS7su25NwqvNJ8JFr71fT/z0+/huyUCyjMvoFBb879U4c9nHkcAEALowAAMlMZ/NsCVEAkAECvL8Xu3d952vfrN9jzf7HBiunrTsNqdwFwBzYzFRWzr2XW9DYb0jUC8a1DIBEJHooqvBzHwX6pEM7sMjBFt3vBCZuqwLk40I28wVqcAAcweisIAKydgbi5L0ilGKSID1Tm4XkYKS/6XWMuvehzP48hU7coVvi2K6lv+xMnjquWLl153eVyEa+88nyzxMSx0smTp6RHRmpt48e/Hvn11+v9xo1LzGvfvqNxxow56UKhkFu4cF7IzJlTI775pkqwOXHimHr27AU32rZtl+JwOIghQ/rHTZ0682aHDo+Z9Ho9lZWVUWeVzKNHD3n98MP3AYsWfaYLCwt3zJw5tVHltlat2ljkcoXr2LEjXvHxPcoAYP/+vZpu3Xp4zvPs2TPKGTNmpcyaNS9t6tTJ4e++Oz6qZ88+Rbt27b/43XdbNYsXL2jUtWv8pTsd/8yZU17r1n1zNTc3W/j66y83S06+LJ8+ffZNlUrNjB49KnrXrh3qoUOHlZw4cVyxYcOa4Pnzl9zQamOsCxbMCZky5f2ItWs36UpKigUffzy1SWLie+k9e/YxbNq0zvfAgX2+3bv3KqkYs3LLlv8Fzp27MCUioqlt1aoVgdOmfRCxbt039ZqNP3r0kGbRouXXZTIZO2HCuMiVKz8LTEyclMtxLPr06V/8xBPdbjIMgxkzPgxfsGBO2JIln6dW9j127Ih6wYKlN4KDQxzjx4+JWrdujf+7707O0emSxStWLG00a9b8G61btzUvXjw/uLS0tFbf1OXLV+mGDx8at3//sXN3clWqjtlsJg8d+knt7x/gidmQSiVsUtKMtOjoZtbk5KuSSZPGR0VFaS29e/fzfKAvXjwv37r1h8upqSnisWNHx3Tv3tMQFRVtW7FiaVB+fp5o69YfLpnNZnLixHFRlX2cTicxefLEpj169C5esWLN9d9//00+Y8aHTRs3bnI1MjLKDgC//PJ/qnnzFt/QamOs48e/HjlmzEsxzz//Uu777ydlLVu2KGjZskWhX3654fqt55GdnUWfPXvau2PHzvqUlBuiuXNnRUyfPiv1scc6l69b96Xf5MkTm27ZsuNKpWJZG1lZGcKkpEmR48ZNyOjb9yl9eXkZmZOTI4yLa2G927PN8/dT3ynldAD/nRiGu1Aiq/rNX3v5azgYJ9Zf2Qy9zf2um8vcmXccNgucjj/nyrgn7UCNZf+WgOLV1SAoGleuXKihNPTvPwQaTdV3+L0vfvUoDQA8SgMAjOgRhVaRPpj0bEt8NTke/TqGQ/IXCtmsxQlXsQVMqRWM0QbW4gTHcuDYmr8lTJkdtgsFsJ7Ng/lYBsp26WDYeAHGjRdhO5cPpsgCQkqDbuQNSYcQKF94BMoXHoG0UxhkXRtBGK6EuJkvxLG+kHUOg7RzGOTdGrtjEXilgYeHpwEkJAwp9PPzdwUFBTtjYpqZoqK05ri4R6xisZjr1OlxQ2rqDSkAPPPM8BKFQsGKRCLuzTfH52ZmZkiMRqPHH6lt2/aGdu06mEmShFgs5iiK4m7eTBGXlZWRKpWKadGiZZ3f08OHD6q7detRHB3dzCaVStnXXnszt/r2bt16luzfv08DAHp9KXXx4nmv/v0HelK+RUfHlD/xRLcygUCA+Pju+vLyMsGYMW/l0zTN9e8/sLS4uEhoMBju6D81YsQL+V5eXmx0dDNbSEiotXXrR8vCwxs7vL29mTZtHjVev66TAsD+/XvV3br1LGnRoqVFJBJx77wzKScl5bosMzNd+PPPR7yDg0Ot/foN0NM0zb300uhCb2+lJ4Bu164dvsOGDc+Pioq2CQQCvPHGuLyMjHRJVlZGvYLIBgwYVBgSEupUqdTMyJEv5h0/flQNAGq1hunb9ymDVCplFQoF++KLo/OuXr2sqN63e/deJU2bRtolEgnXpUt86c2bKVIAOHhwv6p160eNHTt2MolEIm7cuAm5BEHcUQCuDzt3bg/o0aNLy969n2iVnHxVPmXKR2mV2x577PHyZs2aW0mSRGxsc2vnzk+Unjt3tsZYx4x5K1cikXDNm8dZw8IaWa9dS5YAwMmTx9UjR76Yp1KpmZCQUGdCwhCPBevs2dMym81GjRnzVr5QKOQ6d+5S3rp1W8O+fXs0lW3atetoeOSRVpbKZ5umaXbIkGdKKIpCr1599OnpadLq4/joo6SmPXp0afnWW6OjY2PjyseMeStv//4f1a1btzF27fpkGU3T3Kuvvl7gcDjIM2dOyeu6Jj/+uFvdvHmLsoEDB5fSNM2p1RomLq6FFbj7s83z91NfyWoSgI+0Wu15nU6Xfh/H89BCEFU6FkfU1Le+2vseLslonC44hxXx8yGSyDzbti9/D0+/swQk2XC3HxfrAsMxEApCAABx505CPngyQNH43//WgmEYAIBarUG/foM97jgsy2HrkRTPfj5+tT1KjDYU6i1Qe4nROsrtPtWtTUiDx3Q3OI6D5WQWnKn6P70vQkpD0acpH4zMw/MvROHbrqQh1oAHgUbj4xFshUIRq1KpPTMvIpGYtVqtFMMwWLx4QfDJk8dV5eVlNEGQHACUlpYIvL29GQDw8/NzVt/vRx/NSV2/fk3g+vVrQsLCwq2vvz42uy63ltLSElqrjfFsDw4OqeG+89RTCSWjRg1rbjabyb17d6u12hhTQECg55hKparGuBUKhavS9UMikbAAYDabSKVSydR2fB8fX09/oVDEqtWaGtdFry+hK8YpjIrSembH5XI5K5PJmfz8fLq4uJD28fHxjJsgCGg0Gs9yUVGhcPXqz0PXrFnl+RBxHEfk5+fT9XFX8vcP9LQJDg62Gwx6IQBYLBZy/vzZoefOnfUym00CALDZbGT1WJPq91ksFrM2m40EgJKSItrX19ezX5lMxsrliqrZt3tg4MDB+YmJk3KzszOFEyaMi0xLSxXHxja3Am4B/4svlgdnZ2dKXC4X4XS6yI4dH6vx8fT3D/CMVSQSsRaLhQIAvV5PBwYGecZa/f+iogJao/FxVJc7/PwC/r+9+w6vosobOP6dmdvTG0kIhFAPRZQmdlGaDREUURF5XRVXVnGtK4oKIio2UEBsuNj72rBXLIu6olgQGFpoISSkl5ube+/MvH/MTUggDaRzPs+Th9yp50wm5PxODRYUbKud2SUhIbHOdT1mfHx8bT49Hq9ZXR2oF1hOmXLPmh1nsSooKHC2apVae19VVUlKSg7m5eU1OYNMfn6eKz09Y6fZsqD5d1va/1oaONyFPR3rKiHEehqeVan/nkvWAahOud9U6z+2tZpGh80h1mU4yCvawI8fv1hv/6I35jDwgn/u8i2f+P1ZAFI5jjJgW2oir331XzyeX2qDhiOO6EWfPtsf/fz3l7N42fZumRcP6UJGchQZyVFAEntauNCP/5sNmGUN/D/vVHEk+zDLg6AqWIaJFuMmvLX+uAot0YtLJKG4NJwZMShOOZBQkqQD07vv/ifxxx8Xxz/yyLxVmZntgqWlpdqwYYN7WfU68Nevpe7du6+/d+++a0OhkPLCCwtSpk27veN77zU+bWZiYmJo27a82hqTnJzN9WpPWrfOCHXuLCo++eSD+M8//zRp+PAR+Xssg7sgMTEpuHVrbm0TfGVlpVpZWaGlpaWFkpJSQgUF/61Nt2VZFBYW1n5OSkoOXnTRJbkjR47ardrkvLzc2mtt2ZLjio9PCAIsWPBUak7OJs9TTz27IjU1LfzHH797J0y4rLu14wCLBvOTHNq4cYO35rPf71crKsobLCft6ri5Nm0yg1dffd2mBx+8J2vQoKElXq/Xmj59Svuzzx6RP2bMk6s9Ho91773T2paVlbaoXBYfHx/Kzd3iqhlbk5u7/XmkpKSGCgsLXKZp1lZabtuW58rIaNtgYX13JScnh7Kz19U+L/tnXOBKTU1tsqDfqlVqcOXK5VEN7TtQ3m2pcS2tBl8GfAi8BPwX+LOBr0Pc9kdVFpfKtVs0TvB0RDEtOuSEaVNg0GlzmIcXz6k9Lqv/qQBsy1nL2t8XU1XZ4Fi0en7N/4Orv/wXV3/5L1YU2d0Ly7zxAMSXbsUwDCor7YL38OHn1wsagHpBw+hTO+3xVgXLMClfuIqy/6ygfKFOxfurGwwa4sYcQfyYnkQP7Ujsed2IHdmVuFHdiT6tY213o5qvmLO74O6ShCsrXgYNkiQd0Cor/ZrT6bQSEhLDfr9fnT374Yymjg8Gg8rbb7+ZWFpaqjmdTisqKsqsaaVozMCBQ4q++OKzpFWrVnr8fr86f/4T6TseM3ToGYWvv/5K2qZNG7ynnXbWTpV5+8KQIacXffnlZ0l//PG7t7q6Wnn00YcyOnbsXJmZmRU89dRBpTk5m7wfffR+fDgc5tlnn2lVWlpSWxM9fPi521599cX0lSuXewBKS0u1999/N6Hxu9W3cOE7KTk5m53FxUXaiy8+l37SSQOKAfx+v+Zyuc24uHijuLhIe+aZJ1q3PD+nFf/yy09xP/74fXQwGFTmzJnV2rKsBiOEpKSUsKIorF+f3eR4lboGDDi1LCEhMfTqqy+lAAQCAS02Ns7weDzWL78s8X377aLEll7rxBNPLnrppefTi4uLtZyczc733nurdirFPn36VbrdbvPpp+elhUIhZfHib2N+/vmn+DPOOGuPdvkZOvTMol9++Tnu22+/jgmFQsozzzyZ6nQ6rX79jmlyUaRhw4YXLVv2e+zChe8khMNhiooKtT/++N27/br7/92WGteiyFbX9b/t7YQcbN7oMJrR/36UfsAPR8YACq0LDVoXbm/5fT74PRNPPoffvnmXJZ+/Cp/D+dfNQlXtwvEdi++jKNB0l57+8b3RDfD47elSMzOz2LhxPX36HIOperlsxpec0DON80/pxIyXfgFgYJ8Mxg4VeyXfpa/9CaH6Uzd7j26Nu3sKlmGCoqCocgYjSZIOTSNHjipcsuTH2HPPPfOoqKjo8Lhxl2359NOPmpw+77PPPk56/PHZmaZpKunpGYHbbpuS3dTxAwcOKVu1Ss+/7rqrhaoq1oUXjs395puvklwuV+1/vqeddmbJY4890q5//+OKfT5fi+fT39Fdd03OBJgy5Z6Nu3ruySefUr5p04Ytd945qaPfX+no0qVrxd13z1gHkJSUHJ48eerauXMfyZw58/6sAQMGFnXp0rW2QHnGGWeVVFVVqlOnTu5QULDN7fP5jCOP7F02bNg5O/1RfPzxOWnLlv0e89hjT6+u2XbKKYOKrr/+6i7FxcXOo48+puSqqybmAowde2nenXdO6nDWWYN6xccnhM47b/TWJUv+F9+S/HTt2j0wYcK1G++5Z2r76upq7Zxzzt2amJjYYLcpn89njhp1Qe7EiVd2DYcNZcaMh1aHQmFl8uSbO3/xxX+XNnaP0aMv2vrkk/PajhlzybZrrrlu41NPzWvzxBNzMrt161F+wgknFVdUVLSo9uwf//hn7t1335k5evQ5PRMS7FmV3nvv7VQAl8tl3XffQ6sfemhGuzfffC0tISExdOONt2bv6fUjOnfuUn3zzbdlz5kzM3PatNud7dq1r7rvvodWNzUwGuzWl+nTH1j9+OOz2zz66ENZPp/PGDfu8pyacQ576t2W9g6lJc13e5IQogvwHHa/mUJgnK7rq3c4RgNmA6cDFjBD1/X5kX1TgX8ANYPF/qvr+tUtvH0WkF1YWIFp7lq+l778Om90rjOhv2Vx6VP3ArCkVxss1YkaLEc17f9jstMcbEpzcP+JU/hiwX0EIq0N/QZfQMcjTwDg6i//1ej9evi60CNBsObndaztfDxZ2b8yftS5KOr2/1Mum/Flg+c+MOE4kuO8De6ry6wI4v9hM47UaLREL4pLI7S+hLg2cQTiXKi+7d0UzcoQld9swMi3u93GjTsSoziAFuc+KFdOPhBmjtoXZD4PHfszj6qqkJQUDdAee7KM3bJ06a/r27Tp2OyqwVJ9q1at9Fx++SU9vvrq+3qLjY0ceeYR11//rw37YgVlSdqX5Lu9f23evDa5d+9eWQ3ta1GLgxDi9eaO0XV9dAvT8wTwmK7rLwohxgJPAgN3OOZioBPQGTvAWCqE+LzOwOzndV2/qYX322MUy8Kq6deoKLx50dUMe+spwrGZABik4ypYhgIMCZfzbxK45bu7OH3wINJ/yyZ3/XKWLf6Idt368cHG+oX+49OPxqE6OCa9L1mxmTz//FNsYAVru9ndnbq5KlFUze5DWBrgX098X3vuSUem8+3vuQBMPK9ns0GDZVoYBX4qPrIHUIdz6v9e5i9vem2E2NHdURQFR2LzwYkkSZK06z766IP4gQMHl/r9fnXu3Efa9O17dEndoOHDDxfGK4rCiSeeLAtW0iFFvtsHtpYOjm6oGTYB6IrdaqC35CJCiFZAH2BIZNMrwFwhRIqu63VLqxcAT+u6bgLbhBDvAOcDD7YwvXvFjm0UFbHxrOyWVf8Yh5cz3HmodRoEP97wBcTDRQnV5BSX8Z85NwPQM1olocKk1ykjEZEAASAY3Hn8Ut/TL2ywheH+q44jJd7L387s1nz6QwalLy/baburcyJGSQBjmx8txUdC91YUL8vDKKy/9ouzYwK+49vKrkiSJEl/0eOPz0l7881Xdxq70LVr94rHHnt69cKFb6fMnDkjS1VVunXrUT5p0h21XYnGjx8nNm/e7LnllsnZuzNjnyQdqOS7feBr6RiHUxvaLoRoC7wNzGrh/doCObquG5HrGkKILZHtdQOHTGBDnc8bI8fUuFAIMRTYCkzRdf179jITq94icDW2ZB5JyB1FXEkunupKiG6Hu2d/QksXcvu6bTydkUBeZD2BV9rFcUxFAHdkvoGECrvr3q+L3iYhuTWtMgUFBfl8+OE7ABzVuQMrAU/AzzWPLN7p3ndf3p+UeLvW36wKUb0sH8XlwNk2lqolW9CSfVghAzXKRfWyfKzq+jPvOdKjiRrSYafZIRJSYghnxVHTjU2uuixJe5ZhhFFVlXAoBJaJw+Wp/T3bnd87yzIBRf6uHkQmTJi4dcKEiVsb2z9v3vzVje17+unnW1RZJ0kHG/luH/j+0gpZuq5vEkLcBzwALNwzSWrWE8A9uq6HhBBDgHeFEN10XW/xfOCRvrq7ZHNcwwuc5qd3BaA4sS0dV9uF+zdXFnHp4Gvg87lct7GI1S4n/25nTxbxv25u2uYbZG3dPjV0qjOZ1Z//wceOrwG7ZUMBEv58F466llZrVrNNSeLEKA+Xju2DM87Dxud/hQ/X7jwvLhD41f5bFM5teGKDjtceh+pqevxVSkpMk/sPBYdDHuGv5dM0TRQgWB1AURUURSUcrEbVHASrq/CXl1JamI9hhPHFxBEVE4/mcGKEQ1RXVeL2+jAMg6K8HFRVQ9VU+xqhEN6oGKJi4/HFxOFwOnG6PACEQ0GqqyrxV5RhhMNgWfhi4iLHubBME1XTsEyTUCiI0+X+y/msy7IsjHAYzeFosCBuWRbhUJBgoCqSzlJKi7ZhmSaV5aUE/BVgWViWRWVZMf5ye3xTKFRNdZUf0zSoqiyHOuPLFFXF4XDhcLkwQiGCwQBOlxtFUQiHgrjcXlweL26PF6fbi9PlprykkKrKcoKBKizTxLJMnC4PFhaqouLyeHG6PZiGgREO4YmKYcj5V5CY2uQkQJIkSZLUqD2xtK4BtHTOz01AhhBCi7Q2aEDryPa6NgLtgJ8in2tbIHRdr62h0XX9MyHEJuAI4OuWJnh3BkeHWnD42s7H03bDUlzBKj585RO6xxyF//ffiCHE5G0K1lUTKQlVsGTzTxxX/QPlpo/VyT1YHNOFmLJ8ovwuVMNgXcf+hJ0eVoeOBSDJmcDdafYsbbnvrmw8AQ4VZ+sYQhtLQVVwZMSAYdnrJyR6cXVIwJHso7C06UXA5UDTg0soWE1FyTaCgUpU1UFRvt2jweF0Y1SXUrStwB6T4nRhGttbnUzTQFEUAv4K3N4oApVlVFWUojqcYJlUB/xUV5ZjGPtq7R2FnTsENkxzOMECwwihaU58MbGgaISD1YTDQSzTAEXBCIdwub0oqoaCQjhUjeZ0oWoO3B4flmURCgaorqoAyy7Ah4PVWJYdnLjcPkzTJByqxuF02/urA008E8UOZBQF0zDwxSbg8cWgaQ48Li/R8Wk4nC5cHh+q5kDVHGiag1CwilB1AMMIoygKbm8U4VAQ0zRxOJyEQtVohCkvKycUqqaqsgi3N4a45DY43V4URUVzOAkFA5hGGFXVCAUDhKoDqJqGqjkAi7LyEIa66+99ncHRkiRJ0mGspYOjuzew2QV0A+5mewG/Sbqu5wshfgUuAl6M/Lt0h/ENAG8A44UQb2EPjh4BnBRJS4au6zmR73thz5S015u2lB1mn+qiZLPKar/TcZva2TMvre0MiwBfz5M5/6U5lK4pIHV1Ho+ZscCx6EccW++8ipidh5EYTrsb0mBj59YOJdlN7JnCXlzNMEFT0WIbnk46HLILUprLg2WZ9VbBPhBZlkX+ptV4o2KpKC0kuXV7XB4fRjiEoqq16VciBUNVc1CYu56irRspKcghPasbLo8Pty+G2MS0Fq3abZoGlmWhaY3/SoSqqzBNA4fTjWVZBAN+CnOz7XQpKtty1mKaBr6YBLxRscQkpuLxxRAVl1R73VB1FcGAn1ComqiYBBwuD5ZloShghMP4y4pqC6qh6ioUVUVVNYIBP5ZlEfCXYYRDVFWWUZy3idKCLQQDTQeCbp9dE2+EgyhK5PlZFqqmYZoGHl8MJds2o6oa0fHJKIqKYYTxRsfhzeqG2xuF5nCiqg4MI1xb6685nbi90UTHp+BwuqiuqqCqojTyPBQ0hwtFUQiFqolPTkfTnJimgWmaaJEWi0Blae05oeoqiJzndHnwRseiag4URaW6qoJAZRmhYABV1QiHgoRD1fhiEuz8mwEqyitwur1omgPN4dxe6A4GMA0DK9IlyAgHsUyT6qpKLMskNjEVtze69t1zuNw4nC7CwWqqA/YsYk6XByMcst8RhxNPVAwOhwunx4fL7SUuKc0ORrzRdlCzFxxKwa4kSZJ0cGppi8MyGq4KVIAlwBW7cM+rgOeEEHcCxcA4ACHEh8Cduq4vAV4AjgFq+nhO03W9Zt7re4UQfbFbOoLAJXVbIfaWuh0WRmqfkqoUcpS1kvf8J3F22WbyzQ58k+ra6Tx/dCzP/X2y/WEXZiMemBvgy3QPPYvDKMC6qt9Y7V+y/YBC6JFwOt36D0FzNL7+jGka/GeOPQGVoqiRvtAQHZ9CdHwyrdp0ol23o/n5i9e8EyE6AAAgAElEQVRp17UvCamZlG5dyW+LF1FSsIWqihLadTua5NbtKdq6kew/f6i9dqbog9PtpX2PY4hJaMUPHz5PZXkxZYW5xLdqQ2VpIe269iW9fQ9ate1Mbvaf/P7tQqLik4iKSWTdsr0zNCV72fY0OpwuElIziU/JQFFVLMOgOH8zgcoSwmG7u1jNVLmKotjT3VoWqsOBpjkBC0XVthdsd4OiqETFJWGZBpVle279nZjEVFIzBTEJKXhjEoiKScA0DWITU3F5oggGKklvnUJF1b7q9566j+6zM1moliRJkqS9r0XrOAghBjSwOQBsrqn9P0hksZvrOLy18HOWpLUD4CrHK7XbzfJo1JgKnAVZLAu04qu0jgCk5ywn5PJQkNJhp2u137yUuKKtuMoVfHHt6K+2o3LTT1Su/47izB50zBqKo07y1HgPof5uFr/3KOltvbRK97Ax2yJnfSEOp5uk9HZkdu3Hr4veok3no6gOVNKl9wA0h5MvX3+Ezt1j8Po08nICOJwK/kqDUNAkUGUSCtqBhKqCx6cRDlkEq+1tirK9G3Z6Wy/+ijAWEJfgpE073/ZnYFiomsLWzVUUFwYpLQlh/cUlW7xRcbTp0ouyoq0kp7dnW85a8jetJjmjAwU564hJaEXAX05sYiqeqFjyNui06340sQmpxCWnU5CzlmB1JZpaSaAi1+5rXhEiHDIxDAvN4SM2KR2nw8AyK4hLbouJl2p/OZoKLlcl4bBCMKgSNpwEq6pwuLwkt25LOFgFloHT7URzuolNaoeq2t1P4lMyCFVXUR3wU1aYS1VFKf7yYkoLtmKEgySkZuLxxeD0eKmuLCfgL0fVtMjzVvDFJKI5nGgOBy6PD1AIBvy4vVE4nHZNuNsXjcPpblHNdkMFatMIYhp+TCOI5vCBoqEA1f4tmOGa1gsFzelDdUSjam4UxYHq8BAOlmKGq1A1N5ZlYJkhQEHV3Kia3WVG0dyR/NgtPXaLimKPDQgWoygqphFE1dxgmZFuQR4U1YlpVGGZIbtlDAVFc6KqbsLBEoxQBaZZc56FEfajqk4U1UF8Qixl5UYk+HNjGnaQZxlBUDXMsB8zbLdoWKYBloEFqFok2LdMTCMImCiqE1XzYFkGRrAMC9M+BwvLDGMa1aiqM7LYoQsFMMKVkcem1p5v57MaywyiKA57n6JhGlWoqhtFswdEm+EqwqEKLLOamioKRXXVdq9C0bAsA7cLAoGg/XzMICgOVNWFojoB0x5UbRlYZhgL+3tFddr3RUF1+IhLOwnNuetjQeQ6DpIkSYePptZx2OcLwO1nWexm4PD2e5/zU/rOgcOONiRexIo8ldO1Vyip0lhWnkl28hGcqv6AgUq6sg2XYtd0z/yqP1OT7Nn4in59hYKqCpL6peI4ykBxqChlUUT5T8Z9rA8jWELJli8avGf26grycgK1kz6ZkUJ7WoaH9l12vV9ydcDA7WnR4pVNcriTCFfbY9YDwXg8rhIsFJQm+rErigNfQg8C5dmYYT+WFUZ1+OoUaMHpTUVzxqJqbkJV+bijM3G4E6ks/BWHKx5UDSyDqtJVfzkP9RNnX7exdDvcCaiaF4c7AUVzoygaquZBdfjQHFGoDh9YJuFgKZYZRHPG2IVLM0S4usguUGq+SIG75hkpKJoLy6wppNcMnSdS6FexzBDVlZsJBewef5YZihTk3bjcboLBMEF/rl2Y/qsR3a6IdIkCC0Vz2+nfl/ff0xQNRVFRVTcWJlgWplkNFqgOL4riwLLCdkBl1MzHrKI6PJG8W/b7rHkj521/FpojGtXhBZRIoBKO/LzBssL2++VwYpoKiupAUV1YlollBu0gAhUUUFQnCpod1ChqJLizgx6wSM46D5dvpxlAmyUDB2lPGD9+nBg06LTCCy+8uNl34MQT+/V9/vnXlnXo0HGn+clHjDi95w03TFq/LxYHq3uvxx+fk5abm+OeNm3GhubP/GtWr17lnjLl1o75+XnusWMvzbn00ivy9/Y997XZs2e23rIlxz1jxsNNrqQu7Xu7tQCcECIdmAs8pev6J40ccxpwJTBB1/VD7qWur2WBRruiV2gXqQiOjzI5MWotJ7IWsyqKReuTaddt+2xKN5z6P/wl6bgK2+MY5yMNX+Q+kZra2EoqYj+iYn39eyR3uAAjWErx5o8BaN85mvadmw4QUrtcRtCfSzhYgtOTgr/4DzRnDJVFv28/SNHwxQm0inUY4Z1Xpo9LH0iwcjOqMxpfXBc8sZ3rzTpjmWEC5dlUla2isvA3LHN7Xj0ue/6nxoKGmiDDssJUFv1Wb58Z9qM5YzBC5bVBRDhgHwsQCmx/9UKB/EihykBzROOKakNU0lE4nHGEgyWRQnsIj9uioqwQRXXi9LRC1dwYYT9YBka4EqcnBacnBSNUTqgqHyPsjxTQFDRnFIrqQnV4sUwD06jCCFVghMoxQhUEytdhhMqbDDQa1vLBwTufquL0pKJqHtA8ds122E+IasLBAC5fOk5PMnbrgCcSWHgiNeKh2p+Bwx1fmw7TCNi1/GF/pEY+YAdBzpjIs8BuXbDsgq4d6NkBjoVV2xqhAKZlF4Sd7kQU1Y2iOiLPJlKjblbb19c8dh4iz86yDLuFw+FFc8aiOaKwrFBtKwXYrQpxcR6KCouwWwWCdo29okaur6JqHjvQshONotiFa8sI1j56VXXXtkiYkYK/5oxGUV2AaZ/TgJoWlR23YRmRVgZ1p2OtSBCBZUbS2nx3MtkdS5L2r7rT527cuN41ZsyonosW/VBvNe+6cnI2Ox9+eEbm8uXLog3DUBITk0KjRl249fzzL2x2Fsjnn/93Ws+eR5Xfeuudy2HfBks7Gj9+nFi9enWUpqm1f6B69OhZPnv2E2v2dVqk/a+pMQ43AR2AT5s45lPgPuBG4JY9mK4Dzo6Dox1lrQhFlaBowRadH7d2ACNMJ9YfPQnH5FPk9eBN/YZwfC7h+Nz691JdpHS4EEV1kLdqAWDh9KTgSziCuLSTao+LSemPaQSortxM8aaPCAeLd7pvSscxeGM7AeCO2j75VXRSLwCS2o3Y6Zzk5Gjy84vt7hi7QFEdeOM6443rTGLbswCwLAM7ELIwgqVorvhmC0mWZWGEyiM18g0fa1lWpMtGNaojmnB1IaHqQrwxnexxCg1w+dJqv9+lQlh884vrNZQ+RVGwzDBG2I8ZrsQIV2KZBg5XDKojGjNciWlUoahuHK54VIfPLpBbZqRdQYnMzlOFXXPtrVNLrUQK6jWF/sQGC7aHS2EzJjGGgNHQOpXNcPh23qaB5ozaeWMjGnpHFUUBZef/XmuOVRQFRdk7g6ilA4thGGjaX2/BlQ4+U6dObt++fceq//zngz/cbre5YsWf3m3b8lv0i5+fn+c+9dTBe25Q3F901VXXbGxJS9G+EA6HaSxYa4gVmR5bLii3ZzT15IcBM3Vdb7QKVNd1SwjxJHA9h3jgYJr1u1iEY/Pxal0JGPb0qImZ56A5vGxb92rtMSkdLsG/eTnKhjgciXEYBX4US8NZlk5qGRilpxJK2EgwZS0AaeJKwMLpTastYGT2vqPBGs0aqubBG9sJb4+JeyyviqLsctDQ+LVq/mAqONwJLb6/wxXb7DGK5oJIH3WnJzlSo35gqC0gqg47Lw3lp4FtNeMD6lIb2GZv33kwviRJu2fEiNN7nnXWOflffvlZUl5envuEE04qmjjx+py77rq9va6viO7YsXPljBkz1yYkJBg33jixw4oVy2NCoaCSmZlVdfPNt27o2rV7AGDy5JuzXC63mZ+f51qxYnnM1Kn3rAmFQuqTT85tU1RU6PJ4vMaIEeflXXHFVXlNpeepp+alvvvuW6mKAmPH/m3L3Lmz2j3//GvLSktLtMmTb+68cOFnv9UEJB999H78888vaP3KK/9ZPnv2zNYbNmR7nE6XtWTJj/HJySnV06c/sPbzzz9JePfdt1KdTod1442T1g8YMLCsofueeGK/vn//+zUb33nnzdTS0hLnsGEj8s4559zCadNub79p00Zvr159Su+996Fsl8tlAbzyyovJb7zxSlplZYVDiG4Vt95654b09NYhgK+//ip27txZbUtKip0DBgws2rFn9Ouvv5z0xhuvppWWljg7duxcedttd25o27Zdi2rjli9fFjV37qzMkpJiZ//+x5bcfvu0DR6PxyouLtbuvHNS+zVrVkUZhqEI0a1i0qQ7NmRktAmBXXveo8eR5b/9tjR206YN3s6dRcX06fdnJyUlhwHeeuv1xOee+3dGIBDQzjnn3HoTr9TtWnPNNX8XAKedNqA3wP33z1rVr1//yrrHr1u3Juraa2/cFBUVZQL07HlUFVA7y8ann34U98wzT7YpKipyZmVlVd18820bunTpGrjyyku7rFjxZ4yur4x++ul5bfv2PbqksLDQNWXKrZ1VVbUuuODiLaeffmbxmDGjev7znzeuf+GFBRmBQED9v/+7PKd79yMqH3jgnqyCggLXgAGnFt1++10bAbKz17nvu29auw0b1vsUBeuoo/qU3XrrnRvj4+ON7Ox17quu+lu3Bx98dNWRR/by5+ZucV522cXd77hj2rrjjz+pyZqnJ598LG3x4m/jn3nmxZUOh4MXX3w2ZeHCd1o999yry/PztzrHjBnV8+qr/7nhpZeea21ZMHLkqEbf/caeB9i/n2eccfa2RYu+SMzN3eL5/PPvfvntt1+i5s59pO3mzZs9yckpwYkTr9tUk97x48eJ7t2PqPj9999i1q9f53vmmRf/bKjbm7Trmgoc2gHLW3CNFdhjBw5p3si87V2VtbXbHBWp4F2JN3Qk0UlHAdD6iOvZsmwWUYlH4Y1rjzeuPfSwI97gykK0FB9arJuqn3IIrgFXdT/Sel/cZHAgV4OVJOlQ831eSdLSwrK9Gu33TootOC41vsWLg3733TcJjz76+KpwOKxcfvkl3a+//hrfpEm3r+/cWQSuvfaqzi+++GyriROvzz3mmONKp069d73L5bIefvj+NtOm3dHh5Zf/s3z7db5OvOeeB1f369d/TTAYVM47b1jPO+6Ytu7YY4+vKC4u1jZt2tD4VHjAV199HvvOO/9Jmzlzjp6ZmRWcNu2OdrV56t3XHx0dE/766y9jBw4cUgbwyScfJg0aNKQ2nz//vCR+6tTpa6ZPvz/7jjsmZd1007Vdhg49Y9t7733y+5tvvpY0a9aD7QYMGPhHY/dfsuTH2AULXl6+Zctm11VXXdZ9xYpl0VOm3LMuISHRGD9+XNf33ns7cdSoCwq/++6bmOeem5/xwAOPrBaiW9WDD97b5vbbb+nwzDMv6IWFBY67776j4/XX/2v90KFnlLzwwoKUTz/9KGXw4NMKI2mOf/XVl9JnzHh4TYcOnQJPPvlY+p133tphwYKXm1isqN4zSpo5c+6qqKgo84YbJnZ+/PE56ddff/MWyzI544xhBaecMmidYRhMnXpb1oMP3pv5yCPzav94f/31l4kPPvjo6oyMNsFrr/17lwUL5qfedNOkHF1f4XnssUfbTZ/+wOo+ffpVzpr1QEZRUVGDNTRz5z6pjxkzqucnn3y9tLHa706dulTOnHl/u5EjR+X16dO3sk2bzNqgaM2a1e4ZM6Z3mDJl+trjjz+xfMGCp1tNmnRjp1dfffvPp556dtWO40FGjDg9um5XpY0b17vADqBef/29P378cXHMlCmTO/Xq1bt09uwnVoVCIeXyyy/p/v333xUdd9yJFZZlcfHF/7f12GOPLy8vL9NuueWGjvPmzW592213bmrfvkP1pZeO3zx9+pT2zz77yopp0+7IOuWUwYXNBQ0A48dP2Pq///0QN2/e7NZnnTW88Pnn/53x8MNzdI/HUxsmLl36c8xrr727bMOGbPeNN07s0qVLV/+OXa6aeh41QWrNzy0pKTmcn5/nvO22f3X+178mZ5966qDSb7/9Ovauu+7o+NJLbyxLTk4J2+/IF4kPPDBrdadOXQKWZcmC1B7SVLtNFdB0ta8tmjoR9KFKjVSVtFfsterMba3QclPwbOpNdMzRtcc5nDG07nEtCW3OqHe+oii4uyXjSPahuDS8x7clanB7os/sXLtfkiRJ2n9GjDgvv1Wr1HDr1hmhbt26V3TpIip79jyqyuPxWCeccFLJ2rWrfQCjR48pjImJMd1ut/WPf1y7ZePGDd7S0tLa/kj9+h1T0r//sZWqquLxeCxN06x169Z4ysrK1ISEBOPII3s1ufjKF198ljho0JCCrl27B3w+n3nllf/YUnf/oEFDCz/55KMkgOLiIu3333+NHTbsnNpuLV27dis/5ZRBZQ6Hg4EDBxeXl5c5/v73q7c6nU5r2LBzigoKtrlKSkoa7T918cX/tzU2Ntbs2rV7oE2btlV9+hxdlpXVPhgXF2f07Xt06apVug/gk08+TBw0aGjhkUf28rvdbuu6627OWbNmVdTGjetdixZ9GZeR0bbqrLOGFzudTutvfxufHxcXX7ty4nvvvZ1ywQVjtnbp0jXgcDiYMGFi7oYN672bNm1oUVPq8OEj89u0aRtKSEg0xo69NPebb75KBEhMTDLOPPPsEp/PZ8bExJiXXjo+d/nyZfWmEhs8+LTCTp06V3u9XuvkkwcWrVu3xgfw2WefJPTpc3TpccedUOF2u62JE2/YoijKbs8gM2PGzLU9evQsf+ml51pfdNF5PceMOa/70qU/R57dB4l9+vQtHTDg1DKn02ldccVVecFgUF2y5MddmtFk/Ph/5Ho8HmvAgIFlbrfbHDRoSFFKSqvad3jlyhU+gA4dOlYPGHBqmdvttpKTU8Lnn39h3p9//l77XC66aGxBWlp69WWXXdy1pKTYed11N9WbMfOppx5rO2TIyb1qvh555KHWAKqqMmXK3dkffPBuq0mTbuh07rmjt0ZaVuqkcUJuVFSU2b37EVUDBw4p/OyzjxN3zEdLnseIEefltWnTNuT1eq2FC99J6tOnb+mgQUNKVVVlwIBTyzp06Fi5aNGXcTXHDxo0pLBr1+4Bh8OB0+k8rGYC2puaanH4BRgOfNDMNc6JHHtYMVUVqyKEi0xUpX7lkcMV3+z5iqLgzGhJXCZJknToOS41vnBXWgP2haSk5NqCrcvlNhMSEmtneHC7PWZVVZVmGAazZj2Y8d//fpNQXl7mVBR7wGhRUaEjLi7OAGjVqlW9pcXvuuvetc8+Oz/92Wfnt8nMzKq66qprNu/YraWuoqJCpxDdavdnZLSp133n7LNHFI4bd8ERlZWV6ocfLkwUoltFWlp67T3j4xPqpTsmJiZc063J6/WaAJWVFWp8fHyDszfU1NjWPIfExKR6z6W4uNAZSaerSxdRUrMvOjrajIqKNrZu3eosKMh3Jicn16ZbURSSkpJqP2/blu966ql5befPf7J28J1lWcrWrVudLemulJqaXntMRkZGdUlJsQvA7/erDzxwT9ulS3+OrayscAAEAgG17liTuj9nj8djBgIBFaCwcJszJSWl9rpRUVFmdHTM9lk+dlFCQoJx002TcoCcwsICx8yZ97e5445JnRYu/PT3goICZ6tWqbX3UlWVpKTkYF5e3i71E27VKrXOz8Zl1nS5sj+7Tb/frwHk5+c5HnzwvswVK/6MDgSqNMuyiIqKqvfzP/vsEdvuuuv2Tldffd0Gt9tdr6B95ZVXb2psjENmZlawR4+e5UuX/hw3Zsy4nSbKad16+/ublpZevX599k6FpJY8j9TU7e94Xt5W1/ffL04YMuTk2kDBMAylV6/etS0ZrVqltWwQqrRLmgoc5gGvCSEW67r+XEMHCCHGAX8DLtgbiTuQbJ8c0xbdOhYi8Xg4twJ31wOnf70kSZK0d7z77n8Sf/xxcfwjj8xblZnZLlhaWqoNGza4V/2pzevXUvfu3dffu3fftaFQSHnhhQUp06bd3vG99z79nUYkJiaGtm3Lq615z8nZXK8WvnXrjFDnzqLik08+iP/880+Thg8fsV9mNUxMTApu3ZpbW3NWWVmpVlZWaGlpaaGkpJRQQcF/a9NtWRaFhYW1n5OSkoMXXXRJ7siRo3ZrAHBeXm7ttbZsyXHFxycEARYseCo1J2eT56mnnl2RmpoW/uOP370TJlzWvSVTzycmJoc2btzgrfns9/vVioryBstJu9pLICkpOTx27KVbv/76/5KKi4u05OTkUHb2utp72c+nwJWamhpq+Ap/rVfCnDmzMhQF64UXXvszISHR+PjjD+Ife+zRzJr9FRUV6rx5s9sOHDi44KWXnmt9+ulnFickJLZoWsDPP/80btWqlVFHHHFk2cyZ97edOvWeetPVbtmy2dW5swiAXeBPTEzaqUDfkudRt/WnVavU4Mknn1I4bdp9jU6NK3ty7B2NdlXSdf0/wKPAAiHET0KIu4UQ44UQVwghpgkhfgQWALN1XX97XyX4QGEopbXfOzu2bNCvJEmSdHCrrPRrTqfTSkhIDPv9fnX27Iczmjo+GAwqb7/9ZmJpaanmdDqtqKgos6aVojEDBw4p+uKLz5JWrVrp8fv96vz5T+y0+MbQoWcUvv76K2mbNm3wnnbaWSUNXWdvGzLk9KIvv/ws6Y8/fvdWV1crjz76UEbHjp0rMzOzgqeeOqg0J2eT96OP3o8Ph8M8++wzrUpLS2prj4cPP3fbq6++mL5y5XIPQGlpqfb++++2+I/pwoXvpOTkbHYWFxdpL774XPpJJw0oBvD7/ZrL5Tbj4uKN4uIi7Zlnnmjd8vycVvzLLz/F/fjj99HBYFCZM2dW68b6xiclpYQVRWH9+uxGx6s89NB9GStW/OkJh8OUl5erb775WqvU1LTqxMQkY+jQM4t++eXnuG+//TomFAopzzzzZKrT6bT69TumoqFrxcXFh3JyNjU5NqYpVVV+zePxmrGxccaWLTnO1157Oa3u/vvvn962U6fO/mnTZmzo2/fo0nvumdqusWvVVVhY4Jg164F2119/y4a77ro3e8mSH+O++OKzuLrHzJ//ZLrf71dXrlzu+eKLz5IHDx660xSQu/o8hg07p3DJkv/FL1r0RaxhGAQCAWXx4m9jcnI2y+nq9rIm56bSdf1G7K5IZdjTsz4JPAXcDJQD5+i6ftPeTuSBSHPFonjsigjFIaf4kiRJOhyMHDmqMCWlVfW555551Jgx5/Xo0aNno12Oanz22cdJ559/ds/Bg0/s/f7776XcdtuUJhe8GjhwSNnZZ4/Iv+66q8Xo0cOP6N69RyXYXVFqjjnttDNLCgq2ufv3P67Y5/Pt9sqKd901OfOuuyZnNn/kzk4++ZTysWMv3XLnnZM6Dh8+9Kjc3C3uu++esQ7sGvbJk6eunT//iTZnnHFqr5ycTZ4uXbrWFgLPOOOskgsuGJM7derkDoMHn9j7kkvO7/HDD4vjGrrP44/PSbv66vGd62475ZRBRddff3WX0aNH9ExLS6++6qqJuQBjx16aFwxWq2edNajXFVeM69a//7GlDV2zIV27dg9MmHDtxnvumdr+7LOHHhUTExNOTExssLuLz+czR426IHfixCu7Dhlycq+ff/5f1A8/LI4eNOiE3jXHBAIB9fbbb+l02mkDeo8ePbxnfn6e6777HloD0Llzl+qbb74te86cmZlnnjnwqB9+WBx/330Pra4ZCLyjMWMu2frKKy+mDxlycq/5859IbWmealxxxVVb1q1b4xs69OTeN910becTTjiptvD+yScfxi9d+nPcbbdN2QBw8823bVq3bq3v7bffrB2L8MQTczMHDTqhd83X2LHndwOYPn1Ku/79jykZNGhIaWJiknHDDbesnzXr/nZFRYW1Y2h69epdPnr08CNuuGGiGDly1NaGZvTa1efRpk3b0N13z1jz4ovPpZ955sBeI0eeceTLL7+QalmmbGbYy1q8crQQwgEkRT4W6rq+2/3+9qMsdnPl6I/fWMg3WV05U11EpppLm543Y2wNUflFNrHndUONPnSmxjwc5v4/HPIIMp+Hkv2ZR7ly9P61atVKz+WXX9Ljq6++r7fY2MiRZx5x/fX/2rA/FgWTpOa0ZJE86cC0WytH7ygSKDQ57/ThIk1cgerworbxEjfuSNmPTpIkSdqjPvrog/iBAweX+v1+de7cR9r07Xt0Sd3C14cfLoxXFIUTTzxZBg2SJO0zMgTcDQ7X9m6YMmiQJEmSdtXjj89Je/PNV3cau9C1a/eKxx57evXChW+nzJw5I0tVVbp161E+adIdG2uOGT9+nNi8ebPnllsmZ8vVcCVJ2pdk4CBJkiRJ+9iECRO3TpgwcWtj++fNm7+6sX1PP/28vndSJUl7TmZmVvC775b8vL/TIe1Zsqpit8hWBkmSJEmSJOnwIgOHXaQgFx+UJEnaTWZj01tKkiRJ+59pmgrQ6BoeMnDYHfLPniRJ0u5YVl5eEieDB0mSpAOLZVmEQkFnUVFePPBdY8fJMQ4tJNsZJEmS/hrTNK4oKyuaX1ZWdASy4kqSJOlAYiqKUmqaxmzLsh5v7CAZOLSUUjd0kJVlkiRJu6pv3775wPD9nQ5JkiRp9+zzwEEI0QV4DnsxuUJgnK7rq3c4RgNmA6djV/bP0HV9fnP79irZ5CBJkiRJkiQdxvZHU/ETwGO6rncBHgOebOCYi4FOQGfgOGCqECKrBfskSZIkSZIkSdoL9mmLgxCiFdAHGBLZ9AowVwiRouv6tjqHXgA8reu6CWwTQrwDnA882My+5mgAqrrrXY20mCiSvC58aiyqpu7WNQ4mh3r+4PDII8h8Hkr2Vx7r3FfbLwmQJEmSDgj7uqtSWyBH13UDQNd1QwixJbK9buCQCWyo83lj5Jjm9jUnHSAhIWqXE37xJSMi3x2xy+cejJKSovd3Eva6wyGPIPN5KDkA8pgOrN3fiZAkSZL2j8NtcPRPwElALk3MUStJkiTVo2EHDT/t74RIkiRJ+8++Dhw2ARlCCC3S2qABrSibpxMAAAnlSURBVCPb69oItGP7H6m6rQxN7WtONU3MTStJkiQ1SrY0SJIkHeb26eBoXdfzgV+BiyKbLgKW7jC+AeANYLwQQhVCpAAjgDdbsE+SJEmSJEmSpL1gf8yqdBUwUQixCpgY+YwQ4kMhRL/IMS8A64DVwA/ANF3Xs1uwT5IkSZIkSZKkvUCxLLlAgSRJkiRJkiRJTdsfLQ6SJEmSJEmSJB1kZOAgSZIkSZIkSVKzZOAgSZIkSZIkSVKzZOAgSZIkSZIkSVKzDrcF4HabEKIL8ByQBBQC43RdX71/U9U8IUQS9kxUHYEg9mxUf9d1fZsQ4ljgScALrAfGRqbMZXf37W9CiCnAVKCnruvLDrU8CiE8wCxgMBAAvtd1/cqm3s/d3be/CCGGAXcDSuTrLl3X3zrY8yiEeAg4D8gi8n7+lbQfDHmWJEmSDi2yxaHlngAe03W9C/AYdqHyYGABD+i6LnRd74m9iNMMIYQKvAhcHcnTN8AMgN3dt78JIfoAxxJZEPBQzCPwAHbA0CXy87wjsr2p93N39+1zQggFO9C9RNf1XsAlwHORn8nBnsd3gJPZecHKvZGvAyXPkiRJ0iFEBg4tIIRoBfQBXolsegXoE1mA7oCm63qRruuL6mz6AXvl7b5AQNf1mpW0nwBGR77f3X37jRDCjV1AmlBn86GWx2hgHHCHrusWgK7reU29n7u7b9/kqFEmEBf5Ph7IBZI5yPOo6/p3uq5vqrttb/zsDqQ8S5IkSYcWGTi0TFsgR9d1AyDy75bI9oNGpNZ2AvAekEmdmk9d1wsAVQiR+Bf27U/TgBd1XV9fZ9uhlseO2N1OpgghlgghFgkhTqTp93N39+0XkYBoNPCuEGIDdi39uGbSelDlcQd7I18Hep4lSZKkg5QMHA4vc4AKYO7+TsieJIQ4DugHzNvfadnLNKADsFTX9X7ALcBbQPR+TdUeJIRwALcC5+i63g44G3idQyiPkiRJknSwkoFDy2wCMoQQGkDk39aR7QeFyMDMzsAFuq6bwEbsLks1+5MBU9f1or+wb38ZAHQDsoUQ64E2wCdAJw6dPIKdrjCRLii6rv8IFABVNP5+NvXuHojvdS+gta7r/wWI/FuJPa7jUMljXbub9oM5z5IkSdJBSgYOLRCZTedX4KLIpouwa3237b9UtZwQ4l7sfvsjdF2vjmz+GfBGuroAXAW88Rf37Re6rs/Qdb21rutZuq5nAZuB04AHOUTyCLVdpr4ChkDtzDmtgFU08n429e4eoO/1ZqCNEEIACCG6AanYs4EdKnmstbtpP5jzLEmSJB28FMuy9ncaDgpCiK7Y0xsmAMXY0xvq+zdVzRNC9ACWYRcuqyKbs3VdHymEOB57thUP26cczYuct1v7DgSRVodhkelYD6k8CiE6AP/GnmYzBEzWdf2jpt7P3d23vwghLgYmYQ+SBpii6/o7B3sehRCzgXOBNOyWokJd13vsjXwdKHmWJEmSDi0ycJAkSZIkSZIkqVmyq5IkSZIkSZIkSc2SgYMkSZIkSZIkSc2SgYMkSZIkSZIkSc2SgYMkSZIkSZIkSc2SgYMkSZIkSZIkSc2SgYN0SBNCTBVCFOziOaOFEJfupST9JUKIK4UQIxrYvj6yyJ8kSZIkSdJe4djfCZCkA9BoIBl4dj+noyFXYq/L8c4O20cChfs+OZIkSZIkHS5k4CBJe5EQQgHcuq4H9uZ9dF1fujevL0mSJEmSJAMH6bAhhDgF+Ao4FbgaOAPIBx7SdX1e5JhngfMi39esjniXrutTI9vOAe4AjgBKgOexV28ORfZPBa4BRgCzgCOBK4QQbwH3A0OAtkAe8CFwq67rZXXSqAH/Av4GtAO2AZ/run6pEGIR0BfoK4T4v8gpf9N1/dnIatlv6rp+U51rjY6ktUskn89jr8Icjuy/FFgQSeNM4HhgE3Cbrutv1bnOicB9wFGRTeuAe3Rdf6P5py5JkiRJ0qFCjnGQDkdPA79hd+9ZBDwmhOgf2Xc3dnCxFDgu8jUfagvibwH/A4YDd2F3Hbpvh+v7gOci550eOd4HaMBk7IDlDmAgsGPh+8nIdV8HhgE3Rs4F+AewEjvgqEnbBw1lUAgxFHgN+AU4B5gD3ATMbeDwl4H3Is9jNfCqEKJN5DqxwPvYwcJ5wCjgBSC+oftKkiRJknToki0O0uHoFV3XpwNEavHPBs4F/qfr+lohRBGg6rr+Q80JkS5HDwLP67r+jzrbq7EDj/t0Xa8ZY+AFbtB1/d0d7juhznkOIBv4TgiRqev6RiFEV+By4J+6rs+uc95rALquLxdCVALb6qatEdOARbqu17RMfCyEALhPCDFd1/XNdY6dpev6vyPp+hm7NWQY8AR2a0UccI2u6+WR4z9t5t6SJEmSJB2CZIuDdDiqLfhGuhitBto0c04XIBN4XQjhqPkCvgQ82F2XaljARzteQAhxiRBiqRCiAggB39W5NthdqOAvDsqOdHfqw86tGa9h/84ft8P2us+jELtbU83zWAtUAC8LIc4RQsiWBkmSJEk6TMnAQToclezwOYhd+G9KcuTfD7EL/TVf2ZHtbescW6zrerDuyUKIkdhjDL4HzgeOxe4aRJ17JwGVdcc87KZkwIndclBXzefEHbY3+jx0XS/GHpfhxO4+tU0I8YEQosNfTKMkSZIkSQcZ2VVJklqmKPLvldjjH3aUXed7q4H95wM/7tDNacAOxxQCUUKI2L8YPBRgBzWtdtieGvm3iF0Q6RZ1uhDCCwzGHkj9MnbwI0mSJEnSYUIGDpK0s4ZaIHQgB8jSdf3p3bimF6jeYdvFO3z+MvLvOBoexNxY2urRdd2IjFU4H3i8zq7RgInd6rHLdF2vAhYKIY4Abt2da0iSJEmSdPCSgYMk7WwlcE5khebNwBZd17cIIW4EXojMNPQRdiG+A/bUq6N0Xfc3cc3PsAdRTwZ+BM4EBtU9QNd1XQjxFPCwEKIV8A327EWjdF2/sE7aThNCnIbdQpFdZ1B2XVOAT4QQC4BXgZ7YM0Y9vcPA6CYJIc4CLsNecG4jkAH8ne1BjiRJkiRJhwk5xkGSdjYPe8Dwv4GfsLsnoev6a9hTm/bCHnj8FvYUqb9gBxFNeRJ4GPhn5Lx2wJgGjvsH9nSsY7HHUzwC1A1IpgMrsMcb/IQ9I9ROdF3/FLgQ6AcsBK6L3P+aZtK5ozXYXa/uxX4mDwAfYwcTkiRJkiQdRhTLaqg7tiRJkiRJkiRJ0nayxUGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGbJwEGSJEmSJEmSpGb9P/q2DQ+5NiCdAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_group_mean = df_group.groupby(['bandit', 'idx']).median().reset_index()\n",
    "\n",
    "plot_cum_reward(df_group_mean, legend='full', ylim=0.4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Recommendation Metrics and Off-Policy Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We evaluated the bandits according to traditional recommen-\n",
    "dation metrics and off-policy metrics in the test subset of the \"Chicago, USA\" task"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_metrics = [\n",
    "  \"precision_at_1\",\n",
    "  \"ndcg_at_5\",\n",
    "  \"coverage_at_5\",\n",
    "  \"personalization_at_5\",\n",
    "  \"IPS\",\n",
    "  \"SNIPS\",\n",
    "  \"DirectEstimator\",\n",
    "  \"DoublyRobust\"\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/media/workspace/DeepFood/deep-reco-gym/src/mars_gym/tools/eval_viz/util.py:44: FutureWarning:\n",
      "\n",
      "pandas.io.json.json_normalize is deprecated, use pandas.json_normalize instead\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Load Métrics and Params\n",
    "#\n",
    "input_models_eval = fetch_results_path().keys()\n",
    "data_metrics      = load_data_metrics()\n",
    "data_params       = load_eval_params()\n",
    "input_params      = []\n",
    "\n",
    "df_metrics        = filter_df(data_metrics, input_models_eval, input_metrics, \"\")\n",
    "df_eval_params    = filter_df(data_params, input_models_eval, input_params).transpose()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate confidence metrics\n",
    "#\n",
    "confidence_metrics = data_metrics[[c for c in data_metrics.columns if \"_C\" in c]]\n",
    "for c in input_metrics:\n",
    "    c_column = c+\"_C\" \n",
    "    confidence_metrics[c_column] = data_metrics[c_column] if c_column in data_metrics else None\n",
    "confidence_metrics = confidence_metrics[[c+\"_C\" for c in input_metrics]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>precision_at_1</th>\n",
       "      <th>ndcg_at_5</th>\n",
       "      <th>coverage_at_5</th>\n",
       "      <th>personalization_at_5</th>\n",
       "      <th>IPS</th>\n",
       "      <th>SNIPS</th>\n",
       "      <th>DirectEstimator</th>\n",
       "      <th>DoublyRobust</th>\n",
       "      <th>index</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bandit_policy_class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.AdaptiveGreedy</th>\n",
       "      <td>0.317639</td>\n",
       "      <td>0.403579</td>\n",
       "      <td>0.390850</td>\n",
       "      <td>0.767565</td>\n",
       "      <td>0.299046</td>\n",
       "      <td>0.308466</td>\n",
       "      <td>0.200657</td>\n",
       "      <td>0.267138</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.CustomRewardModelLinUCB</th>\n",
       "      <td>0.328249</td>\n",
       "      <td>0.442917</td>\n",
       "      <td>0.362862</td>\n",
       "      <td>0.729376</td>\n",
       "      <td>0.305576</td>\n",
       "      <td>0.315562</td>\n",
       "      <td>0.199502</td>\n",
       "      <td>0.266484</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.EpsilonGreedy</th>\n",
       "      <td>0.302056</td>\n",
       "      <td>0.443252</td>\n",
       "      <td>0.343287</td>\n",
       "      <td>0.733547</td>\n",
       "      <td>0.296847</td>\n",
       "      <td>0.295397</td>\n",
       "      <td>0.187270</td>\n",
       "      <td>0.254888</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.ExploreThenExploit</th>\n",
       "      <td>0.307692</td>\n",
       "      <td>0.418607</td>\n",
       "      <td>0.333071</td>\n",
       "      <td>0.731641</td>\n",
       "      <td>0.296997</td>\n",
       "      <td>0.294139</td>\n",
       "      <td>0.190652</td>\n",
       "      <td>0.256225</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.FixedPolicy</th>\n",
       "      <td>0.074271</td>\n",
       "      <td>0.171294</td>\n",
       "      <td>0.373850</td>\n",
       "      <td>0.759891</td>\n",
       "      <td>0.076116</td>\n",
       "      <td>0.077219</td>\n",
       "      <td>0.084967</td>\n",
       "      <td>0.077790</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.LinThompsonSampling</th>\n",
       "      <td>0.039788</td>\n",
       "      <td>0.137056</td>\n",
       "      <td>0.423900</td>\n",
       "      <td>0.771211</td>\n",
       "      <td>0.037423</td>\n",
       "      <td>0.035373</td>\n",
       "      <td>0.041869</td>\n",
       "      <td>0.039049</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.LinUCB</th>\n",
       "      <td>0.075928</td>\n",
       "      <td>0.207314</td>\n",
       "      <td>0.271450</td>\n",
       "      <td>0.696228</td>\n",
       "      <td>0.053439</td>\n",
       "      <td>0.056094</td>\n",
       "      <td>0.054993</td>\n",
       "      <td>0.051205</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.PercentileAdaptiveGreedy</th>\n",
       "      <td>0.336870</td>\n",
       "      <td>0.438648</td>\n",
       "      <td>0.375563</td>\n",
       "      <td>0.743905</td>\n",
       "      <td>0.321505</td>\n",
       "      <td>0.317082</td>\n",
       "      <td>0.198404</td>\n",
       "      <td>0.280774</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.RandomPolicy</th>\n",
       "      <td>0.039788</td>\n",
       "      <td>0.138404</td>\n",
       "      <td>0.389600</td>\n",
       "      <td>0.775653</td>\n",
       "      <td>0.041457</td>\n",
       "      <td>0.040577</td>\n",
       "      <td>0.042541</td>\n",
       "      <td>0.041919</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mars_gym.model.bandit.SoftmaxExplorer</th>\n",
       "      <td>0.301724</td>\n",
       "      <td>0.452869</td>\n",
       "      <td>0.330988</td>\n",
       "      <td>0.725512</td>\n",
       "      <td>0.287223</td>\n",
       "      <td>0.287718</td>\n",
       "      <td>0.188744</td>\n",
       "      <td>0.253449</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                precision_at_1  ndcg_at_5  \\\n",
       "bandit_policy_class                                                         \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.317639        0.403579    \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.328249        0.442917    \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.302056        0.443252    \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.307692        0.418607    \n",
       "mars_gym.model.bandit.FixedPolicy               0.074271        0.171294    \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.039788        0.137056    \n",
       "mars_gym.model.bandit.LinUCB                    0.075928        0.207314    \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.336870        0.438648    \n",
       "mars_gym.model.bandit.RandomPolicy              0.039788        0.138404    \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.301724        0.452869    \n",
       "\n",
       "                                                coverage_at_5  \\\n",
       "bandit_policy_class                                             \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.390850        \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.362862        \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.343287        \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.333071        \n",
       "mars_gym.model.bandit.FixedPolicy               0.373850        \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.423900        \n",
       "mars_gym.model.bandit.LinUCB                    0.271450        \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.375563        \n",
       "mars_gym.model.bandit.RandomPolicy              0.389600        \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.330988        \n",
       "\n",
       "                                                personalization_at_5  \\\n",
       "bandit_policy_class                                                    \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.767565               \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.729376               \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.733547               \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.731641               \n",
       "mars_gym.model.bandit.FixedPolicy               0.759891               \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.771211               \n",
       "mars_gym.model.bandit.LinUCB                    0.696228               \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.743905               \n",
       "mars_gym.model.bandit.RandomPolicy              0.775653               \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.725512               \n",
       "\n",
       "                                                     IPS     SNIPS  \\\n",
       "bandit_policy_class                                                  \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.299046  0.308466   \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.305576  0.315562   \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.296847  0.295397   \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.296997  0.294139   \n",
       "mars_gym.model.bandit.FixedPolicy               0.076116  0.077219   \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.037423  0.035373   \n",
       "mars_gym.model.bandit.LinUCB                    0.053439  0.056094   \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.321505  0.317082   \n",
       "mars_gym.model.bandit.RandomPolicy              0.041457  0.040577   \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.287223  0.287718   \n",
       "\n",
       "                                                DirectEstimator  DoublyRobust  \\\n",
       "bandit_policy_class                                                             \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0.200657         0.267138       \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0.199502         0.266484       \n",
       "mars_gym.model.bandit.EpsilonGreedy             0.187270         0.254888       \n",
       "mars_gym.model.bandit.ExploreThenExploit        0.190652         0.256225       \n",
       "mars_gym.model.bandit.FixedPolicy               0.084967         0.077790       \n",
       "mars_gym.model.bandit.LinThompsonSampling       0.041869         0.039049       \n",
       "mars_gym.model.bandit.LinUCB                    0.054993         0.051205       \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0.198404         0.280774       \n",
       "mars_gym.model.bandit.RandomPolicy              0.042541         0.041919       \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0.188744         0.253449       \n",
       "\n",
       "                                                index  \n",
       "bandit_policy_class                                    \n",
       "mars_gym.model.bandit.AdaptiveGreedy            0      \n",
       "mars_gym.model.bandit.CustomRewardModelLinUCB   0      \n",
       "mars_gym.model.bandit.EpsilonGreedy             0      \n",
       "mars_gym.model.bandit.ExploreThenExploit        0      \n",
       "mars_gym.model.bandit.FixedPolicy               0      \n",
       "mars_gym.model.bandit.LinThompsonSampling       0      \n",
       "mars_gym.model.bandit.LinUCB                    0      \n",
       "mars_gym.model.bandit.PercentileAdaptiveGreedy  0      \n",
       "mars_gym.model.bandit.RandomPolicy              0      \n",
       "mars_gym.model.bandit.SoftmaxExplorer           0      "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Merge and Mean \n",
    "#\n",
    "df_metrics['model_task'] = data_metrics.model_task\n",
    "df_params['model_task']  = df_params.iteraction.apply(lambda x: x.split(\"/\")[-1])\n",
    "\n",
    "df_result = df_metrics.merge(\n",
    "    df_params[['bandit_policy_class', 'model_task']].reset_index()\n",
    ", on='model_task')\n",
    "\n",
    "df_result = df_result.groupby(['bandit_policy_class']).mean()\n",
    "df_result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_result.round(3).to_csv('metrics.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# fig = plot_bar(df_result, confidence = None, title=\"Comparison of Recsys Metrics\")\n",
    "\n",
    "# fig.update_layout(xaxis_showgrid=False, yaxis_showgrid=False)\n",
    "# fig.update_layout(font={'family': 'Courier New, monospace', 'size': 14}, height=550)\n",
    "# fig#.show(renderer=\"svg\")\n",
    "\n",
    "# #fig.write_image(\"output/tmp/metrics/recsys_rank.eps\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fairness Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We evaluated the SoftmaxExplorer bandit on the “RecSys Cities\" task, in the perspective of\n",
    "disparate mistreatment and disparate treatment. We selected a few attributes that we judged to be sensitive for all partners in the marketplace."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# task_id: InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature - Accessibility"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_models_eval = 'InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee'\n",
    "list_input_features    = ['accessible parking', 'accessible hotel']\n",
    "input_metrics     = 'true_positive_rate'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>accessible parking.1</td>\n",
       "      <td>accessible parking</td>\n",
       "      <td>1063</td>\n",
       "      <td>2380</td>\n",
       "      <td>0.222664</td>\n",
       "      <td>0.023616</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>accessible parking.0</td>\n",
       "      <td>accessible parking</td>\n",
       "      <td>3193</td>\n",
       "      <td>10038</td>\n",
       "      <td>0.297201</td>\n",
       "      <td>0.013806</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>accessible hotel.1</td>\n",
       "      <td>accessible hotel</td>\n",
       "      <td>1345</td>\n",
       "      <td>3799</td>\n",
       "      <td>0.242414</td>\n",
       "      <td>0.020492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>accessible hotel.0</td>\n",
       "      <td>accessible hotel</td>\n",
       "      <td>3114</td>\n",
       "      <td>8619</td>\n",
       "      <td>0.287845</td>\n",
       "      <td>0.014361</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  feature             sub_key  total_class  total_individuals  \\\n",
       "sub                                                                             \n",
       "1    accessible parking.1  accessible parking  1063         2380                \n",
       "0    accessible parking.0  accessible parking  3193         10038               \n",
       "1    accessible hotel.1    accessible hotel    1345         3799                \n",
       "0    accessible hotel.0    accessible hotel    3114         8619                \n",
       "\n",
       "     true_positive_rate  true_positive_rate_C  \n",
       "sub                                            \n",
       "1    0.222664            0.023616              \n",
       "0    0.297201            0.013806              \n",
       "1    0.242414            0.020492              \n",
       "0    0.287845            0.014361              "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# fig, df_metrics, df_total = plot_fairness_metrics(input_models_eval, input_features, input_metrics, sub_map = None)\n",
    "metrics = []\n",
    "sub_map = None\n",
    "\n",
    "for input_features in list_input_features:\n",
    "    df_all_metrics    = load_fairness_metrics().loc[input_models_eval]\n",
    "    df_instances      = load_fairness_df().loc[input_models_eval]\n",
    "    df_all_metric_filter = df_all_metrics[df_all_metrics.sub_key.isin([input_features])]\n",
    "\n",
    "    columns         = list(np.unique(['sub_key', 'sub', 'feature', 'total_class', 'total_individuals'] + [input_metrics]))\n",
    "    if input_metrics+\"_C\" in df_all_metrics.columns:\n",
    "        columns.append(input_metrics+\"_C\")\n",
    "\n",
    "    df_metrics      = filter_df(df_all_metrics, input_models_eval, columns, 'sub')\n",
    "\n",
    "    df_metrics      = df_metrics[df_metrics.sub_key.isin([input_features])]\n",
    "    if sub_map != None:\n",
    "        df_metrics['sub'] = df_metrics['sub'].map(sub_map)\n",
    "\n",
    "    df_metrics      = df_metrics.sort_values(\"feature\").set_index(\"sub\")\n",
    "    df_metrics      = df_metrics.sort_values(input_metrics)\n",
    "    metrics.append(df_metrics)\n",
    "\n",
    "metrics = pd.concat(metrics)\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "      <th>color</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>parking&lt;br&gt;True</th>\n",
       "      <td>accessible parking.1</td>\n",
       "      <td>accessible parking</td>\n",
       "      <td>1063</td>\n",
       "      <td>2380</td>\n",
       "      <td>0.222664</td>\n",
       "      <td>0.023616</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>parking&lt;br&gt;False</th>\n",
       "      <td>accessible parking.0</td>\n",
       "      <td>accessible parking</td>\n",
       "      <td>3193</td>\n",
       "      <td>10038</td>\n",
       "      <td>0.297201</td>\n",
       "      <td>0.013806</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hotel&lt;br&gt;True</th>\n",
       "      <td>accessible hotel.1</td>\n",
       "      <td>accessible hotel</td>\n",
       "      <td>1345</td>\n",
       "      <td>3799</td>\n",
       "      <td>0.242414</td>\n",
       "      <td>0.020492</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hotel&lt;br&gt;False</th>\n",
       "      <td>accessible hotel.0</td>\n",
       "      <td>accessible hotel</td>\n",
       "      <td>3114</td>\n",
       "      <td>8619</td>\n",
       "      <td>0.287845</td>\n",
       "      <td>0.014361</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                               feature             sub_key  total_class  \\\n",
       "sub                                                                       \n",
       "parking<br>True   accessible parking.1  accessible parking  1063          \n",
       "parking<br>False  accessible parking.0  accessible parking  3193          \n",
       "hotel<br>True     accessible hotel.1    accessible hotel    1345          \n",
       "hotel<br>False    accessible hotel.0    accessible hotel    3114          \n",
       "\n",
       "                  total_individuals  true_positive_rate  true_positive_rate_C  \\\n",
       "sub                                                                             \n",
       "parking<br>True   2380               0.222664            0.023616               \n",
       "parking<br>False  10038              0.297201            0.013806               \n",
       "hotel<br>True     3799               0.242414            0.020492               \n",
       "hotel<br>False    8619               0.287845            0.014361               \n",
       "\n",
       "                  color  \n",
       "sub                      \n",
       "parking<br>True   2      \n",
       "parking<br>False  2      \n",
       "hotel<br>True     5      \n",
       "hotel<br>False    5      "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics.reset_index()\n",
    "metrics['sub'] = ['parking<br>True', 'parking<br>False', 'hotel<br>True', 'hotel<br>False']\n",
    "metrics['color'] = [2,2, 5,5]\n",
    "metrics = metrics.set_index('sub')\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "error_x": {
          "array": [
           0.02361613479151095,
           0.0138059237113749,
           0.020492447312361814,
           0.014361164579029424
          ],
          "type": "data"
         },
         "marker": {
          "color": [
           "rgb(103, 219, 165)",
           "rgb(103, 219, 165)",
           "rgb(44, 152, 160)",
           "rgb(44, 152, 160)"
          ]
         },
         "orientation": "h",
         "type": "bar",
         "x": [
          0.22266417712722056,
          0.2972006460716394,
          0.2424144599186233,
          0.28784461570988074
         ],
         "y": [
          "parking<br>True",
          "parking<br>False",
          "hotel<br>True",
          "hotel<br>False"
         ]
        }
       ],
       "layout": {
        "font": {
         "family": "Courier New, monospace",
         "size": 22
        },
        "height": 550,
        "legend": {
         "orientation": "h",
         "y": -0.2
        },
        "shapes": [
         {
          "line": {
           "dash": "dot",
           "width": 1
          },
          "type": "line",
          "x0": 0,
          "x1": 1,
          "xref": "paper",
          "y0": 1.5,
          "y1": 1.5,
          "yref": "y"
         }
        ],
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "white",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "#C8D4E3"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "white",
          "polar": {
           "angularaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           },
           "bgcolor": "white",
           "radialaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "yaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "zaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "bgcolor": "white",
           "caxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          }
         }
        },
        "title": {
         "text": ""
        },
        "xaxis": {
         "nticks": 4,
         "showgrid": false,
         "title": {
          "text": "true_positive_rate"
         }
        },
        "yaxis": {
         "categoryarray": [
          "parking<br>False",
          "parking<br>True",
          "hotel<br>False",
          "hotel<br>True"
         ],
         "categoryorder": "array",
         "showgrid": false,
         "title": {
          "text": "Accessible"
         }
        }
       }
      },
      "text/html": [
       "<div>\n",
       "        \n",
       "        \n",
       "            <div id=\"53ecedfb-d19f-4f3f-b69e-c8667562872d\" class=\"plotly-graph-div\" style=\"height:550px; width:100%;\"></div>\n",
       "            <script type=\"text/javascript\">\n",
       "                require([\"plotly\"], function(Plotly) {\n",
       "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
       "                    \n",
       "                if (document.getElementById(\"53ecedfb-d19f-4f3f-b69e-c8667562872d\")) {\n",
       "                    Plotly.newPlot(\n",
       "                        '53ecedfb-d19f-4f3f-b69e-c8667562872d',\n",
       "                        [{\"error_x\": {\"array\": [0.02361613479151095, 0.0138059237113749, 0.020492447312361814, 0.014361164579029424], \"type\": \"data\"}, \"marker\": {\"color\": [\"rgb(103, 219, 165)\", \"rgb(103, 219, 165)\", \"rgb(44, 152, 160)\", \"rgb(44, 152, 160)\"]}, \"orientation\": \"h\", \"type\": \"bar\", \"x\": [0.22266417712722056, 0.2972006460716394, 0.2424144599186233, 0.28784461570988074], \"y\": [\"parking<br>True\", \"parking<br>False\", \"hotel<br>True\", \"hotel<br>False\"]}],\n",
       "                        {\"font\": {\"family\": \"Courier New, monospace\", \"size\": 22}, \"height\": 550, \"legend\": {\"orientation\": \"h\", \"y\": -0.2}, \"shapes\": [{\"line\": {\"dash\": \"dot\", \"width\": 1}, \"type\": \"line\", \"x0\": 0, \"x1\": 1, \"xref\": \"paper\", \"y0\": 1.5, \"y1\": 1.5, \"yref\": \"y\"}], \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"white\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"#C8D4E3\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"white\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"radialaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"yaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"zaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"caxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"\"}, \"xaxis\": {\"nticks\": 4, \"showgrid\": false, \"title\": {\"text\": \"true_positive_rate\"}}, \"yaxis\": {\"categoryarray\": [\"parking<br>False\", \"parking<br>True\", \"hotel<br>False\", \"hotel<br>True\"], \"categoryorder\": \"array\", \"showgrid\": false, \"title\": {\"text\": \"Accessible\"}}},\n",
       "                        {\"responsive\": true}\n",
       "                    ).then(function(){\n",
       "                            \n",
       "var gd = document.getElementById('53ecedfb-d19f-4f3f-b69e-c8667562872d');\n",
       "var x = new MutationObserver(function (mutations, observer) {{\n",
       "        var display = window.getComputedStyle(gd).display;\n",
       "        if (!display || display === 'none') {{\n",
       "            console.log([gd, 'removed!']);\n",
       "            Plotly.purge(gd);\n",
       "            observer.disconnect();\n",
       "        }}\n",
       "}});\n",
       "\n",
       "// Listen for the removal of the full notebook cells\n",
       "var notebookContainer = gd.closest('#notebook-container');\n",
       "if (notebookContainer) {{\n",
       "    x.observe(notebookContainer, {childList: true});\n",
       "}}\n",
       "\n",
       "// Listen for the clearing of the current output cell\n",
       "var outputEl = gd.closest('.output');\n",
       "if (outputEl) {{\n",
       "    x.observe(outputEl, {childList: true});\n",
       "}}\n",
       "\n",
       "                        })\n",
       "                };\n",
       "                });\n",
       "            </script>\n",
       "        </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plot_fairness_mistreatment(metrics, input_metrics, title=\"\")\n",
    "\n",
    "fig.update_layout(shapes=[\n",
    "  dict(\n",
    "    type= 'line',\n",
    "    line=dict(\n",
    "        width=1,\n",
    "        dash=\"dot\",\n",
    "    ),        \n",
    "    xref='paper', x0= 0, x1= 1,\n",
    "    yref='y',     y0= 1.5, y1= 1.5\n",
    "  )\n",
    "])\n",
    "\n",
    "# Set custom x-axis labels\n",
    "fig.update_xaxes(nticks=4)\n",
    "\n",
    "fig.update_layout(yaxis={'categoryorder':'array', 'categoryarray':['parking<br>False', 'parking<br>True', 'hotel<br>False', 'hotel<br>True']})\n",
    "fig.update_layout(xaxis_showgrid=False, yaxis_showgrid=False, yaxis_title=\"Accessible\")\n",
    "fig.update_layout(font={'family': 'Courier New, monospace', 'size': 22}, height=550)\n",
    "\n",
    "fig.show()\n",
    "#fig.write_image(\"output/tmp/metrics/paper/recsys_fairnes_{}.eps\".format(\"accessible\"))    \n",
    "#fig.show(renderer=\"svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature - Accommodation Type"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_models_eval = 'InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee'\n",
    "list_input_features    = ['hotel', 'house / apartment']\n",
    "input_metrics     = 'true_positive_rate'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sub</th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>hotel.0</td>\n",
       "      <td>hotel</td>\n",
       "      <td>1649</td>\n",
       "      <td>2111</td>\n",
       "      <td>0.205584</td>\n",
       "      <td>0.020963</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>hotel.1</td>\n",
       "      <td>hotel</td>\n",
       "      <td>2579</td>\n",
       "      <td>10307</td>\n",
       "      <td>0.302903</td>\n",
       "      <td>0.013879</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>house / apartment.0</td>\n",
       "      <td>house / apartment</td>\n",
       "      <td>3044</td>\n",
       "      <td>11727</td>\n",
       "      <td>0.307458</td>\n",
       "      <td>0.013011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>house / apartment.1</td>\n",
       "      <td>house / apartment</td>\n",
       "      <td>743</td>\n",
       "      <td>691</td>\n",
       "      <td>0.162758</td>\n",
       "      <td>0.030811</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  sub              feature            sub_key  total_class  total_individuals  \\\n",
       "0  0   hotel.0              hotel              1649         2111                \n",
       "1  1   hotel.1              hotel              2579         10307               \n",
       "3  0   house / apartment.0  house / apartment  3044         11727               \n",
       "2  1   house / apartment.1  house / apartment  743          691                 \n",
       "\n",
       "   true_positive_rate  true_positive_rate_C  \n",
       "0  0.205584            0.020963              \n",
       "1  0.302903            0.013879              \n",
       "3  0.307458            0.013011              \n",
       "2  0.162758            0.030811              "
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics = []\n",
    "for input_features in list_input_features:\n",
    "    df_all_metrics    = load_fairness_metrics().loc[input_models_eval]\n",
    "    df_instances      = load_fairness_df().loc[input_models_eval]\n",
    "    df_all_metric_filter = df_all_metrics[df_all_metrics.sub_key.isin([input_features])]\n",
    "\n",
    "    columns         = list(np.unique(['sub_key', 'sub', 'feature', 'total_class', 'total_individuals'] + [input_metrics]))\n",
    "    if input_metrics+\"_C\" in df_all_metrics.columns:\n",
    "        columns.append(input_metrics+\"_C\")\n",
    "\n",
    "    df_metrics      = filter_df(df_all_metrics, input_models_eval, columns, 'sub')\n",
    "\n",
    "    df_metrics      = df_metrics[df_metrics.sub_key.isin([input_features])]\n",
    "    if sub_map != None:\n",
    "        df_metrics['sub'] = df_metrics['sub'].map(sub_map)\n",
    "\n",
    "    df_metrics      = df_metrics.sort_values(\"feature\").set_index(\"sub\")\n",
    "    df_metrics      = df_metrics.sort_values(input_metrics)\n",
    "    metrics.append(df_metrics)\n",
    "\n",
    "metrics = pd.concat(metrics).reset_index().sort_values(['sub_key', 'sub'])\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "      <th>color</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>hotel&lt;br&gt;False</th>\n",
       "      <td>hotel.0</td>\n",
       "      <td>hotel</td>\n",
       "      <td>1649</td>\n",
       "      <td>2111</td>\n",
       "      <td>0.205584</td>\n",
       "      <td>0.020963</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hotel&lt;br&gt;True</th>\n",
       "      <td>hotel.1</td>\n",
       "      <td>hotel</td>\n",
       "      <td>2579</td>\n",
       "      <td>10307</td>\n",
       "      <td>0.302903</td>\n",
       "      <td>0.013879</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>house/apart&lt;br&gt;False</th>\n",
       "      <td>house / apartment.0</td>\n",
       "      <td>house / apartment</td>\n",
       "      <td>3044</td>\n",
       "      <td>11727</td>\n",
       "      <td>0.307458</td>\n",
       "      <td>0.013011</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>house/apart&lt;br&gt;True</th>\n",
       "      <td>house / apartment.1</td>\n",
       "      <td>house / apartment</td>\n",
       "      <td>743</td>\n",
       "      <td>691</td>\n",
       "      <td>0.162758</td>\n",
       "      <td>0.030811</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  feature            sub_key  total_class  \\\n",
       "sub                                                                         \n",
       "hotel<br>False        hotel.0              hotel              1649          \n",
       "hotel<br>True         hotel.1              hotel              2579          \n",
       "house/apart<br>False  house / apartment.0  house / apartment  3044          \n",
       "house/apart<br>True   house / apartment.1  house / apartment  743           \n",
       "\n",
       "                      total_individuals  true_positive_rate  \\\n",
       "sub                                                           \n",
       "hotel<br>False        2111               0.205584             \n",
       "hotel<br>True         10307              0.302903             \n",
       "house/apart<br>False  11727              0.307458             \n",
       "house/apart<br>True   691                0.162758             \n",
       "\n",
       "                      true_positive_rate_C  color  \n",
       "sub                                                \n",
       "hotel<br>False        0.020963              5      \n",
       "hotel<br>True         0.013879              5      \n",
       "house/apart<br>False  0.013011              2      \n",
       "house/apart<br>True   0.030811              2      "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#metrics.reset_index().sort_values('sub_key')\n",
    "metrics['color'] = [5,5, 2,2]\n",
    "metrics['sub']   = ['hotel<br>False', 'hotel<br>True', 'house/apart<br>False', 'house/apart<br>True']\n",
    "metrics = metrics.set_index('sub')\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "error_x": {
          "array": [
           0.020962783439803054,
           0.013878558663566494,
           0.013010644821110588,
           0.030811013633295792
          ],
          "type": "data"
         },
         "marker": {
          "color": [
           "rgb(44, 152, 160)",
           "rgb(44, 152, 160)",
           "rgb(103, 219, 165)",
           "rgb(103, 219, 165)"
          ]
         },
         "orientation": "h",
         "type": "bar",
         "x": [
          0.20558372218744236,
          0.3029033688947088,
          0.30745821671009826,
          0.16275784666589266
         ],
         "y": [
          "hotel<br>False",
          "hotel<br>True",
          "house/apart<br>False",
          "house/apart<br>True"
         ]
        }
       ],
       "layout": {
        "font": {
         "family": "Courier New, monospace",
         "size": 22
        },
        "height": 550,
        "legend": {
         "orientation": "h",
         "y": -0.2
        },
        "shapes": [
         {
          "line": {
           "dash": "dot",
           "width": 1
          },
          "type": "line",
          "x0": 0,
          "x1": 1,
          "xref": "paper",
          "y0": 1.5,
          "y1": 1.5,
          "yref": "y"
         }
        ],
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "white",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "#C8D4E3"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "white",
          "polar": {
           "angularaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           },
           "bgcolor": "white",
           "radialaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "yaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "zaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "bgcolor": "white",
           "caxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          }
         }
        },
        "title": {
         "text": ""
        },
        "xaxis": {
         "nticks": 4,
         "showgrid": false,
         "title": {
          "text": "true_positive_rate"
         }
        },
        "yaxis": {
         "categoryarray": [
          "house/apart<br>False",
          "house/apart<br>True",
          "hotel<br>False",
          "hotel<br>True"
         ],
         "categoryorder": "array",
         "showgrid": false,
         "title": {
          "text": "Business"
         }
        }
       }
      },
      "text/html": [
       "<div>\n",
       "        \n",
       "        \n",
       "            <div id=\"949fb933-4a6c-4310-b6b5-1fe26016f27b\" class=\"plotly-graph-div\" style=\"height:550px; width:100%;\"></div>\n",
       "            <script type=\"text/javascript\">\n",
       "                require([\"plotly\"], function(Plotly) {\n",
       "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
       "                    \n",
       "                if (document.getElementById(\"949fb933-4a6c-4310-b6b5-1fe26016f27b\")) {\n",
       "                    Plotly.newPlot(\n",
       "                        '949fb933-4a6c-4310-b6b5-1fe26016f27b',\n",
       "                        [{\"error_x\": {\"array\": [0.020962783439803054, 0.013878558663566494, 0.013010644821110588, 0.030811013633295792], \"type\": \"data\"}, \"marker\": {\"color\": [\"rgb(44, 152, 160)\", \"rgb(44, 152, 160)\", \"rgb(103, 219, 165)\", \"rgb(103, 219, 165)\"]}, \"orientation\": \"h\", \"type\": \"bar\", \"x\": [0.20558372218744236, 0.3029033688947088, 0.30745821671009826, 0.16275784666589266], \"y\": [\"hotel<br>False\", \"hotel<br>True\", \"house/apart<br>False\", \"house/apart<br>True\"]}],\n",
       "                        {\"font\": {\"family\": \"Courier New, monospace\", \"size\": 22}, \"height\": 550, \"legend\": {\"orientation\": \"h\", \"y\": -0.2}, \"shapes\": [{\"line\": {\"dash\": \"dot\", \"width\": 1}, \"type\": \"line\", \"x0\": 0, \"x1\": 1, \"xref\": \"paper\", \"y0\": 1.5, \"y1\": 1.5, \"yref\": \"y\"}], \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"white\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"#C8D4E3\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"white\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"radialaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"yaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"zaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"caxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"\"}, \"xaxis\": {\"nticks\": 4, \"showgrid\": false, \"title\": {\"text\": \"true_positive_rate\"}}, \"yaxis\": {\"categoryarray\": [\"house/apart<br>False\", \"house/apart<br>True\", \"hotel<br>False\", \"hotel<br>True\"], \"categoryorder\": \"array\", \"showgrid\": false, \"title\": {\"text\": \"Business\"}}},\n",
       "                        {\"responsive\": true}\n",
       "                    ).then(function(){\n",
       "                            \n",
       "var gd = document.getElementById('949fb933-4a6c-4310-b6b5-1fe26016f27b');\n",
       "var x = new MutationObserver(function (mutations, observer) {{\n",
       "        var display = window.getComputedStyle(gd).display;\n",
       "        if (!display || display === 'none') {{\n",
       "            console.log([gd, 'removed!']);\n",
       "            Plotly.purge(gd);\n",
       "            observer.disconnect();\n",
       "        }}\n",
       "}});\n",
       "\n",
       "// Listen for the removal of the full notebook cells\n",
       "var notebookContainer = gd.closest('#notebook-container');\n",
       "if (notebookContainer) {{\n",
       "    x.observe(notebookContainer, {childList: true});\n",
       "}}\n",
       "\n",
       "// Listen for the clearing of the current output cell\n",
       "var outputEl = gd.closest('.output');\n",
       "if (outputEl) {{\n",
       "    x.observe(outputEl, {childList: true});\n",
       "}}\n",
       "\n",
       "                        })\n",
       "                };\n",
       "                });\n",
       "            </script>\n",
       "        </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plot_fairness_mistreatment(metrics, input_metrics, title=\"\")\n",
    "\n",
    "fig.update_layout(shapes=[\n",
    "  dict(\n",
    "    type= 'line',\n",
    "    line=dict(\n",
    "        width=1,\n",
    "        dash=\"dot\",\n",
    "    ),        \n",
    "    xref='paper', x0= 0, x1= 1,\n",
    "    yref='y',     y0= 1.5, y1= 1.5\n",
    "  )\n",
    "])\n",
    "\n",
    "# Set custom x-axis labels\n",
    "fig.update_xaxes(nticks=4)\n",
    "\n",
    "fig.update_layout(yaxis={'categoryorder':'array', 'categoryarray':['house/apart<br>False', 'house/apart<br>True', 'hotel<br>False', 'hotel<br>True']})\n",
    "fig.update_layout(xaxis_showgrid=False, yaxis_showgrid=False, yaxis_title=\"Business\")\n",
    "fig.update_layout(font={'family': 'Courier New, monospace', 'size': 22}, height=550)\n",
    "\n",
    "\n",
    "fig.show()\n",
    "#fig.write_image(\"output/tmp/metrics/paper/recsys_fairnes_{}.eps\".format(\"business\"))    \n",
    "#fig.show(renderer=\"svg\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature - Business Affinity\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_models_eval = 'InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee'\n",
    "list_input_features    = ['childcare', 'family friendly']\n",
    "input_metrics     = 'true_positive_rate'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>childcare.1</td>\n",
       "      <td>childcare</td>\n",
       "      <td>1007</td>\n",
       "      <td>2490</td>\n",
       "      <td>0.237465</td>\n",
       "      <td>0.024281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>childcare.0</td>\n",
       "      <td>childcare</td>\n",
       "      <td>3258</td>\n",
       "      <td>9928</td>\n",
       "      <td>0.287760</td>\n",
       "      <td>0.013512</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>family friendly.0</td>\n",
       "      <td>family friendly</td>\n",
       "      <td>2113</td>\n",
       "      <td>3313</td>\n",
       "      <td>0.200593</td>\n",
       "      <td>0.017856</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>family friendly.1</td>\n",
       "      <td>family friendly</td>\n",
       "      <td>2384</td>\n",
       "      <td>9105</td>\n",
       "      <td>0.311986</td>\n",
       "      <td>0.014863</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               feature          sub_key  total_class  total_individuals  \\\n",
       "sub                                                                       \n",
       "1    childcare.1        childcare        1007         2490                \n",
       "0    childcare.0        childcare        3258         9928                \n",
       "0    family friendly.0  family friendly  2113         3313                \n",
       "1    family friendly.1  family friendly  2384         9105                \n",
       "\n",
       "     true_positive_rate  true_positive_rate_C  \n",
       "sub                                            \n",
       "1    0.237465            0.024281              \n",
       "0    0.287760            0.013512              \n",
       "0    0.200593            0.017856              \n",
       "1    0.311986            0.014863              "
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics = []\n",
    "for input_features in list_input_features:\n",
    "    df_all_metrics    = load_fairness_metrics().loc[input_models_eval]\n",
    "    df_instances      = load_fairness_df().loc[input_models_eval]\n",
    "    df_all_metric_filter = df_all_metrics[df_all_metrics.sub_key.isin([input_features])]\n",
    "\n",
    "    columns         = list(np.unique(['sub_key', 'sub', 'feature', 'total_class', 'total_individuals'] + [input_metrics]))\n",
    "    if input_metrics+\"_C\" in df_all_metrics.columns:\n",
    "        columns.append(input_metrics+\"_C\")\n",
    "\n",
    "    df_metrics      = filter_df(df_all_metrics, input_models_eval, columns, 'sub')\n",
    "\n",
    "    df_metrics      = df_metrics[df_metrics.sub_key.isin([input_features])]\n",
    "    if sub_map != None:\n",
    "        df_metrics['sub'] = df_metrics['sub'].map(sub_map)\n",
    "\n",
    "    df_metrics      = df_metrics.sort_values(\"feature\").set_index(\"sub\")\n",
    "    df_metrics      = df_metrics.sort_values(input_metrics)\n",
    "    metrics.append(df_metrics)\n",
    "\n",
    "metrics = pd.concat(metrics)\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "      <th>color</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>childcare&lt;br&gt;True</th>\n",
       "      <td>childcare.1</td>\n",
       "      <td>childcare</td>\n",
       "      <td>1007</td>\n",
       "      <td>2490</td>\n",
       "      <td>0.237465</td>\n",
       "      <td>0.024281</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>childcare&lt;br&gt;False</th>\n",
       "      <td>childcare.0</td>\n",
       "      <td>childcare</td>\n",
       "      <td>3258</td>\n",
       "      <td>9928</td>\n",
       "      <td>0.287760</td>\n",
       "      <td>0.013512</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>family friendly&lt;br&gt;True</th>\n",
       "      <td>family friendly.0</td>\n",
       "      <td>family friendly</td>\n",
       "      <td>2113</td>\n",
       "      <td>3313</td>\n",
       "      <td>0.200593</td>\n",
       "      <td>0.017856</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>family friendly&lt;br&gt;False</th>\n",
       "      <td>family friendly.1</td>\n",
       "      <td>family friendly</td>\n",
       "      <td>2384</td>\n",
       "      <td>9105</td>\n",
       "      <td>0.311986</td>\n",
       "      <td>0.014863</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    feature          sub_key  total_class  \\\n",
       "sub                                                                         \n",
       "childcare<br>True         childcare.1        childcare        1007          \n",
       "childcare<br>False        childcare.0        childcare        3258          \n",
       "family friendly<br>True   family friendly.0  family friendly  2113          \n",
       "family friendly<br>False  family friendly.1  family friendly  2384          \n",
       "\n",
       "                          total_individuals  true_positive_rate  \\\n",
       "sub                                                               \n",
       "childcare<br>True         2490               0.237465             \n",
       "childcare<br>False        9928               0.287760             \n",
       "family friendly<br>True   3313               0.200593             \n",
       "family friendly<br>False  9105               0.311986             \n",
       "\n",
       "                          true_positive_rate_C  color  \n",
       "sub                                                    \n",
       "childcare<br>True         0.024281              2      \n",
       "childcare<br>False        0.013512              2      \n",
       "family friendly<br>True   0.017856              5      \n",
       "family friendly<br>False  0.014863              5      "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics.reset_index()\n",
    "metrics['sub'] = ['childcare<br>True', 'childcare<br>False', 'family friendly<br>True', 'family friendly<br>False']\n",
    "metrics['color'] = [2,2, 5,5]\n",
    "metrics = metrics.set_index('sub')\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "error_x": {
          "array": [
           0.024280997706685955,
           0.013512349576754215,
           0.017856184951342592,
           0.01486349284279912
          ],
          "type": "data"
         },
         "marker": {
          "color": [
           "rgb(103, 219, 165)",
           "rgb(103, 219, 165)",
           "rgb(44, 152, 160)",
           "rgb(44, 152, 160)"
          ]
         },
         "orientation": "h",
         "type": "bar",
         "x": [
          0.23746490387655814,
          0.287760222726847,
          0.2005926436392602,
          0.31198632092779505
         ],
         "y": [
          "childcare<br>True",
          "childcare<br>False",
          "family friendly<br>True",
          "family friendly<br>False"
         ]
        }
       ],
       "layout": {
        "font": {
         "family": "Courier New, monospace",
         "size": 22
        },
        "height": 550,
        "legend": {
         "orientation": "h",
         "y": -0.2
        },
        "shapes": [
         {
          "line": {
           "dash": "dot",
           "width": 1
          },
          "type": "line",
          "x0": 0,
          "x1": 1,
          "xref": "paper",
          "y0": 1.5,
          "y1": 1.5,
          "yref": "y"
         }
        ],
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "white",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "#C8D4E3"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "white",
          "polar": {
           "angularaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           },
           "bgcolor": "white",
           "radialaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "yaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "zaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "bgcolor": "white",
           "caxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          }
         }
        },
        "title": {
         "text": ""
        },
        "xaxis": {
         "nticks": 4,
         "showgrid": false,
         "title": {
          "text": "true_positive_rate"
         }
        },
        "yaxis": {
         "categoryarray": [
          "childcare<br>False",
          "childcare<br>True",
          "family friendly<br>False",
          "family friendly<br>True"
         ],
         "categoryorder": "array",
         "showgrid": false,
         "title": {
          "text": "Business"
         }
        }
       }
      },
      "text/html": [
       "<div>\n",
       "        \n",
       "        \n",
       "            <div id=\"3fad4293-8cc9-4529-a100-1e6a3ddc518a\" class=\"plotly-graph-div\" style=\"height:550px; width:100%;\"></div>\n",
       "            <script type=\"text/javascript\">\n",
       "                require([\"plotly\"], function(Plotly) {\n",
       "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
       "                    \n",
       "                if (document.getElementById(\"3fad4293-8cc9-4529-a100-1e6a3ddc518a\")) {\n",
       "                    Plotly.newPlot(\n",
       "                        '3fad4293-8cc9-4529-a100-1e6a3ddc518a',\n",
       "                        [{\"error_x\": {\"array\": [0.024280997706685955, 0.013512349576754215, 0.017856184951342592, 0.01486349284279912], \"type\": \"data\"}, \"marker\": {\"color\": [\"rgb(103, 219, 165)\", \"rgb(103, 219, 165)\", \"rgb(44, 152, 160)\", \"rgb(44, 152, 160)\"]}, \"orientation\": \"h\", \"type\": \"bar\", \"x\": [0.23746490387655814, 0.287760222726847, 0.2005926436392602, 0.31198632092779505], \"y\": [\"childcare<br>True\", \"childcare<br>False\", \"family friendly<br>True\", \"family friendly<br>False\"]}],\n",
       "                        {\"font\": {\"family\": \"Courier New, monospace\", \"size\": 22}, \"height\": 550, \"legend\": {\"orientation\": \"h\", \"y\": -0.2}, \"shapes\": [{\"line\": {\"dash\": \"dot\", \"width\": 1}, \"type\": \"line\", \"x0\": 0, \"x1\": 1, \"xref\": \"paper\", \"y0\": 1.5, \"y1\": 1.5, \"yref\": \"y\"}], \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"white\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"#C8D4E3\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"white\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"radialaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"yaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"zaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"caxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"\"}, \"xaxis\": {\"nticks\": 4, \"showgrid\": false, \"title\": {\"text\": \"true_positive_rate\"}}, \"yaxis\": {\"categoryarray\": [\"childcare<br>False\", \"childcare<br>True\", \"family friendly<br>False\", \"family friendly<br>True\"], \"categoryorder\": \"array\", \"showgrid\": false, \"title\": {\"text\": \"Business\"}}},\n",
       "                        {\"responsive\": true}\n",
       "                    ).then(function(){\n",
       "                            \n",
       "var gd = document.getElementById('3fad4293-8cc9-4529-a100-1e6a3ddc518a');\n",
       "var x = new MutationObserver(function (mutations, observer) {{\n",
       "        var display = window.getComputedStyle(gd).display;\n",
       "        if (!display || display === 'none') {{\n",
       "            console.log([gd, 'removed!']);\n",
       "            Plotly.purge(gd);\n",
       "            observer.disconnect();\n",
       "        }}\n",
       "}});\n",
       "\n",
       "// Listen for the removal of the full notebook cells\n",
       "var notebookContainer = gd.closest('#notebook-container');\n",
       "if (notebookContainer) {{\n",
       "    x.observe(notebookContainer, {childList: true});\n",
       "}}\n",
       "\n",
       "// Listen for the clearing of the current output cell\n",
       "var outputEl = gd.closest('.output');\n",
       "if (outputEl) {{\n",
       "    x.observe(outputEl, {childList: true});\n",
       "}}\n",
       "\n",
       "                        })\n",
       "                };\n",
       "                });\n",
       "            </script>\n",
       "        </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plot_fairness_mistreatment(metrics, input_metrics, title=\"\")\n",
    "\n",
    "\n",
    "\n",
    "fig.update_layout(shapes=[\n",
    "  dict(\n",
    "    type= 'line',\n",
    "    line=dict(\n",
    "        width=1,\n",
    "        dash=\"dot\",\n",
    "    ),        \n",
    "    xref='paper', x0= 0, x1= 1,\n",
    "    yref='y',     y0= 1.5, y1= 1.5\n",
    "  )\n",
    "])\n",
    "\n",
    "# Set custom x-axis labels\n",
    "fig.update_xaxes(nticks=4)\n",
    "\n",
    "fig.update_layout(yaxis={'categoryorder':'array', 'categoryarray':['childcare<br>False', 'childcare<br>True', 'family friendly<br>False', 'family friendly<br>True']})\n",
    "fig.update_layout(xaxis_showgrid=False, yaxis_showgrid=False, yaxis_title=\"Business\")\n",
    "fig.update_layout(font={'family': 'Courier New, monospace', 'size': 22}, height=550)\n",
    "\n",
    "\n",
    "\n",
    "fig.show()\n",
    "#fig.write_image(\"output/tmp/metrics/paper/recsys_fairnes_{}.eps\".format(\"businesscare\"))    \n",
    "#fig.show(renderer=\"svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature - City"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_models_eval = 'InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee'\n",
    "input_features    = 'city_idx'\n",
    "input_metrics     = 'true_positive_rate'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "sub_map = {'0': \"Boston\",\n",
    "            '1': \"Como\",\n",
    "            '2': \"Chicago\",\n",
    "            '3': \"Lausanne\",\n",
    "            '4': \"Dublin\",\n",
    "            '5': \"Rio de Janeiro\",\n",
    "            '6': \"Barcelona\",\n",
    "            '7': \"Copenhagen\",\n",
    "            '8': \"Vienna\",\n",
    "            '9': \"New York\",\n",
    "            '10': \"Hong Kong\",\n",
    "            '11': \"Vancouver\"}\n",
    "\n",
    "fig, df_metrics, df_total = plot_fairness_metrics(input_models_eval, input_features, input_metrics, sub_map = sub_map)            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "error_x": {
          "array": [
           0.16483524552302928,
           0.041650477069622764,
           0.05727762786965144,
           0.037876047322174033,
           0.02993755970856302,
           0.03161661057003146,
           0.0824746955900816,
           0.08552886526732101,
           0.039507683631567536,
           0.0647105694127374,
           0.02675609571268609,
           0.24002335665827296
          ],
          "type": "data"
         },
         "marker": {
          "color": [
           0,
           1,
           2,
           3,
           4,
           5,
           6,
           7,
           8,
           9,
           10,
           11
          ],
          "colorscale": [
           [
            0,
            "rgb(176, 242, 188)"
           ],
           [
            0.16666666666666666,
            "rgb(137, 232, 172)"
           ],
           [
            0.3333333333333333,
            "rgb(103, 219, 165)"
           ],
           [
            0.5,
            "rgb(76, 200, 163)"
           ],
           [
            0.6666666666666666,
            "rgb(56, 178, 163)"
           ],
           [
            0.8333333333333334,
            "rgb(44, 152, 160)"
           ],
           [
            1,
            "rgb(37, 125, 152)"
           ]
          ]
         },
         "orientation": "h",
         "type": "bar",
         "x": [
          0.18421052631578946,
          0.2529075343063283,
          0.28952801108788506,
          0.29110487187293177,
          0.2916910982734073,
          0.29718205314688123,
          0.3090088383838384,
          0.3132959734329597,
          0.3236781450429776,
          0.3349471861471861,
          0.3360587371957656,
          0.4509803921568627
         ],
         "y": [
          "Como",
          "Dublin",
          "Copenhagen",
          "Vienna",
          "Barcelona",
          "Rio de Janeiro",
          "Boston",
          "Vancouver",
          "Hong Kong",
          "Chicago",
          "New York",
          "Lausanne"
         ]
        }
       ],
       "layout": {
        "font": {
         "family": "Courier New, monospace",
         "size": 17
        },
        "height": 550,
        "legend": {
         "orientation": "h",
         "y": -0.2
        },
        "shapes": [
         {
          "line": {
           "dash": "dot",
           "width": 1
          },
          "type": "line",
          "x0": 0.3062161139469011,
          "x1": 0.3062161139469011,
          "xref": "x",
          "y0": 0,
          "y1": 1,
          "yref": "paper"
         }
        ],
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "white",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "#C8D4E3"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "white",
          "polar": {
           "angularaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           },
           "bgcolor": "white",
           "radialaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "yaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "zaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "bgcolor": "white",
           "caxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          }
         }
        },
        "title": {
         "text": ""
        },
        "xaxis": {
         "showgrid": false,
         "title": {
          "text": "true_positive_rate"
         }
        },
        "yaxis": {
         "showgrid": false,
         "title": {
          "text": "city_idx"
         }
        }
       }
      },
      "text/html": [
       "<div>\n",
       "        \n",
       "        \n",
       "            <div id=\"508f3d7e-fe65-45eb-8f76-3c0c5c7c8767\" class=\"plotly-graph-div\" style=\"height:550px; width:100%;\"></div>\n",
       "            <script type=\"text/javascript\">\n",
       "                require([\"plotly\"], function(Plotly) {\n",
       "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
       "                    \n",
       "                if (document.getElementById(\"508f3d7e-fe65-45eb-8f76-3c0c5c7c8767\")) {\n",
       "                    Plotly.newPlot(\n",
       "                        '508f3d7e-fe65-45eb-8f76-3c0c5c7c8767',\n",
       "                        [{\"error_x\": {\"array\": [0.16483524552302928, 0.041650477069622764, 0.05727762786965144, 0.037876047322174033, 0.02993755970856302, 0.03161661057003146, 0.0824746955900816, 0.08552886526732101, 0.039507683631567536, 0.0647105694127374, 0.02675609571268609, 0.24002335665827296], \"type\": \"data\"}, \"marker\": {\"color\": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], \"colorscale\": [[0.0, \"rgb(176, 242, 188)\"], [0.16666666666666666, \"rgb(137, 232, 172)\"], [0.3333333333333333, \"rgb(103, 219, 165)\"], [0.5, \"rgb(76, 200, 163)\"], [0.6666666666666666, \"rgb(56, 178, 163)\"], [0.8333333333333334, \"rgb(44, 152, 160)\"], [1.0, \"rgb(37, 125, 152)\"]]}, \"orientation\": \"h\", \"type\": \"bar\", \"x\": [0.18421052631578946, 0.2529075343063283, 0.28952801108788506, 0.29110487187293177, 0.2916910982734073, 0.29718205314688123, 0.3090088383838384, 0.3132959734329597, 0.3236781450429776, 0.3349471861471861, 0.3360587371957656, 0.4509803921568627], \"y\": [\"Como\", \"Dublin\", \"Copenhagen\", \"Vienna\", \"Barcelona\", \"Rio de Janeiro\", \"Boston\", \"Vancouver\", \"Hong Kong\", \"Chicago\", \"New York\", \"Lausanne\"]}],\n",
       "                        {\"font\": {\"family\": \"Courier New, monospace\", \"size\": 17}, \"height\": 550, \"legend\": {\"orientation\": \"h\", \"y\": -0.2}, \"shapes\": [{\"line\": {\"dash\": \"dot\", \"width\": 1}, \"type\": \"line\", \"x0\": 0.3062161139469011, \"x1\": 0.3062161139469011, \"xref\": \"x\", \"y0\": 0, \"y1\": 1, \"yref\": \"paper\"}], \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"white\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"#C8D4E3\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"white\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"radialaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"yaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"zaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"caxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"\"}, \"xaxis\": {\"showgrid\": false, \"title\": {\"text\": \"true_positive_rate\"}}, \"yaxis\": {\"showgrid\": false, \"title\": {\"text\": \"city_idx\"}}},\n",
       "                        {\"responsive\": true}\n",
       "                    ).then(function(){\n",
       "                            \n",
       "var gd = document.getElementById('508f3d7e-fe65-45eb-8f76-3c0c5c7c8767');\n",
       "var x = new MutationObserver(function (mutations, observer) {{\n",
       "        var display = window.getComputedStyle(gd).display;\n",
       "        if (!display || display === 'none') {{\n",
       "            console.log([gd, 'removed!']);\n",
       "            Plotly.purge(gd);\n",
       "            observer.disconnect();\n",
       "        }}\n",
       "}});\n",
       "\n",
       "// Listen for the removal of the full notebook cells\n",
       "var notebookContainer = gd.closest('#notebook-container');\n",
       "if (notebookContainer) {{\n",
       "    x.observe(notebookContainer, {childList: true});\n",
       "}}\n",
       "\n",
       "// Listen for the clearing of the current output cell\n",
       "var outputEl = gd.closest('.output');\n",
       "if (outputEl) {{\n",
       "    x.observe(outputEl, {childList: true});\n",
       "}}\n",
       "\n",
       "                        })\n",
       "                };\n",
       "                });\n",
       "            </script>\n",
       "        </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig.show()\n",
    "#fig.write_image(\"output/tmp/metrics/recsys_fairnes_{}.eps\".format(input_features))    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>feature</th>\n",
       "      <th>sub_key</th>\n",
       "      <th>total_class</th>\n",
       "      <th>total_individuals</th>\n",
       "      <th>true_positive_rate</th>\n",
       "      <th>true_positive_rate_C</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sub</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Como</th>\n",
       "      <td>city_idx.1</td>\n",
       "      <td>city_idx</td>\n",
       "      <td>34</td>\n",
       "      <td>30</td>\n",
       "      <td>0.184211</td>\n",
       "      <td>0.164835</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Dublin</th>\n",
       "      <td>city_idx.4</td>\n",
       "      <td>city_idx</td>\n",
       "      <td>265</td>\n",
       "      <td>1035</td>\n",
       "      <td>0.252908</td>\n",
       "      <td>0.041650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Copenhagen</th>\n",
       "      <td>city_idx.7</td>\n",
       "      <td>city_idx</td>\n",
       "      <td>127</td>\n",
       "      <td>572</td>\n",
       "      <td>0.289528</td>\n",
       "      <td>0.057278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Vienna</th>\n",
       "      <td>city_idx.8</td>\n",
       "      <td>city_idx</td>\n",
       "      <td>379</td>\n",
       "      <td>1282</td>\n",
       "      <td>0.291105</td>\n",
       "      <td>0.037876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Barcelona</th>\n",
       "      <td>city_idx.6</td>\n",
       "      <td>city_idx</td>\n",
       "      <td>640</td>\n",
       "      <td>1652</td>\n",
       "      <td>0.291691</td>\n",
       "      <td>0.029938</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               feature   sub_key  total_class  total_individuals  \\\n",
       "sub                                                                \n",
       "Como        city_idx.1  city_idx  34           30                  \n",
       "Dublin      city_idx.4  city_idx  265          1035                \n",
       "Copenhagen  city_idx.7  city_idx  127          572                 \n",
       "Vienna      city_idx.8  city_idx  379          1282                \n",
       "Barcelona   city_idx.6  city_idx  640          1652                \n",
       "\n",
       "            true_positive_rate  true_positive_rate_C  \n",
       "sub                                                   \n",
       "Como        0.184211            0.164835              \n",
       "Dublin      0.252908            0.041650              \n",
       "Copenhagen  0.289528            0.057278              \n",
       "Vienna      0.291105            0.037876              \n",
       "Barcelona   0.291691            0.029938              "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_metrics.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feature Device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_models_eval = 'InteractionTraining____mars_gym_model_b___logit_multipli_9dd8714dfd_96b9d686ee'\n",
    "df_instances      = load_fairness_df().loc[input_models_eval]\n",
    "\n",
    "input_features    = 'device_idx'\n",
    "input_metrics     = 'true_positive_rate'\n",
    "input_items_top = True\n",
    "input_items     = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/marlesson/anaconda3/lib/python3.7/site-packages/numpy/core/_methods.py:234: RuntimeWarning:\n",
      "\n",
      "Degrees of freedom <= 0 for slice\n",
      "\n",
      "/home/marlesson/anaconda3/lib/python3.7/site-packages/numpy/core/_methods.py:226: RuntimeWarning:\n",
      "\n",
      "invalid value encountered in double_scalars\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "plotlyServerURL": "https://plot.ly"
       },
       "data": [
        {
         "error_y": {
          "array": [
           0.014036470388890544,
           0.060212925815169425,
           0.10332698899204755,
           0.0959358085637333,
           0.07321121520045266
          ],
          "type": "data"
         },
         "name": "Desktop",
         "type": "bar",
         "x": [
          "ID:371",
          "ID:6085",
          "ID:367",
          "ID:1628",
          "ID:253"
         ],
         "y": [
          0.06323620701829592,
          0.22016996845151438,
          0.142281240497071,
          0.14953048378229142,
          0.3343912796236766
         ]
        },
        {
         "error_y": {
          "array": [
           0.11677800265583932,
           0.055632390434544036,
           0.029213313471479974,
           0.07462909997845024,
           0.08317211170852465
          ],
          "type": "data"
         },
         "name": "Mobile",
         "type": "bar",
         "x": [
          "ID:371",
          "ID:6085",
          "ID:367",
          "ID:1628",
          "ID:253"
         ],
         "y": [
          0.2983235341355043,
          0.24468733149546165,
          0.0761359004182336,
          0.15907974237842218,
          0.3029204527071367
         ]
        },
        {
         "error_y": {
          "array": [
           0.18454737007101824,
           0.19135757350732704,
           0.17778834011917766,
           0.030325356749184643,
           0.1608608491907772
          ],
          "type": "data"
         },
         "name": "Tablet",
         "type": "bar",
         "x": [
          "ID:371",
          "ID:6085",
          "ID:367",
          "ID:1628",
          "ID:253"
         ],
         "y": [
          0.17655574902892113,
          0.33074622314709884,
          0.13608611188828945,
          0.09330697770097426,
          0.34038441663142294
         ]
        }
       ],
       "layout": {
        "font": {
         "family": "Courier New, monospace",
         "size": 21
        },
        "height": 600,
        "legend": {
         "orientation": "h",
         "y": -0.2
        },
        "shapes": [
         {
          "line": {
           "dash": "dot",
           "width": 1
          },
          "type": "line",
          "x0": 0,
          "x1": 1,
          "xref": "paper",
          "y0": 0.20452237459362096,
          "y1": 0.20452237459362096,
          "yref": "y"
         }
        ],
        "template": {
         "data": {
          "bar": [
           {
            "error_x": {
             "color": "#2a3f5f"
            },
            "error_y": {
             "color": "#2a3f5f"
            },
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "bar"
           }
          ],
          "barpolar": [
           {
            "marker": {
             "line": {
              "color": "white",
              "width": 0.5
             }
            },
            "type": "barpolar"
           }
          ],
          "carpet": [
           {
            "aaxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "baxis": {
             "endlinecolor": "#2a3f5f",
             "gridcolor": "#C8D4E3",
             "linecolor": "#C8D4E3",
             "minorgridcolor": "#C8D4E3",
             "startlinecolor": "#2a3f5f"
            },
            "type": "carpet"
           }
          ],
          "choropleth": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "choropleth"
           }
          ],
          "contour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "contour"
           }
          ],
          "contourcarpet": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "contourcarpet"
           }
          ],
          "heatmap": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmap"
           }
          ],
          "heatmapgl": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "heatmapgl"
           }
          ],
          "histogram": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "histogram"
           }
          ],
          "histogram2d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2d"
           }
          ],
          "histogram2dcontour": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "histogram2dcontour"
           }
          ],
          "mesh3d": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "type": "mesh3d"
           }
          ],
          "parcoords": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "parcoords"
           }
          ],
          "pie": [
           {
            "automargin": true,
            "type": "pie"
           }
          ],
          "scatter": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter"
           }
          ],
          "scatter3d": [
           {
            "line": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatter3d"
           }
          ],
          "scattercarpet": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattercarpet"
           }
          ],
          "scattergeo": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergeo"
           }
          ],
          "scattergl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattergl"
           }
          ],
          "scattermapbox": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scattermapbox"
           }
          ],
          "scatterpolar": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolar"
           }
          ],
          "scatterpolargl": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterpolargl"
           }
          ],
          "scatterternary": [
           {
            "marker": {
             "colorbar": {
              "outlinewidth": 0,
              "ticks": ""
             }
            },
            "type": "scatterternary"
           }
          ],
          "surface": [
           {
            "colorbar": {
             "outlinewidth": 0,
             "ticks": ""
            },
            "colorscale": [
             [
              0,
              "#0d0887"
             ],
             [
              0.1111111111111111,
              "#46039f"
             ],
             [
              0.2222222222222222,
              "#7201a8"
             ],
             [
              0.3333333333333333,
              "#9c179e"
             ],
             [
              0.4444444444444444,
              "#bd3786"
             ],
             [
              0.5555555555555556,
              "#d8576b"
             ],
             [
              0.6666666666666666,
              "#ed7953"
             ],
             [
              0.7777777777777778,
              "#fb9f3a"
             ],
             [
              0.8888888888888888,
              "#fdca26"
             ],
             [
              1,
              "#f0f921"
             ]
            ],
            "type": "surface"
           }
          ],
          "table": [
           {
            "cells": {
             "fill": {
              "color": "#EBF0F8"
             },
             "line": {
              "color": "white"
             }
            },
            "header": {
             "fill": {
              "color": "#C8D4E3"
             },
             "line": {
              "color": "white"
             }
            },
            "type": "table"
           }
          ]
         },
         "layout": {
          "annotationdefaults": {
           "arrowcolor": "#2a3f5f",
           "arrowhead": 0,
           "arrowwidth": 1
          },
          "coloraxis": {
           "colorbar": {
            "outlinewidth": 0,
            "ticks": ""
           }
          },
          "colorscale": {
           "diverging": [
            [
             0,
             "#8e0152"
            ],
            [
             0.1,
             "#c51b7d"
            ],
            [
             0.2,
             "#de77ae"
            ],
            [
             0.3,
             "#f1b6da"
            ],
            [
             0.4,
             "#fde0ef"
            ],
            [
             0.5,
             "#f7f7f7"
            ],
            [
             0.6,
             "#e6f5d0"
            ],
            [
             0.7,
             "#b8e186"
            ],
            [
             0.8,
             "#7fbc41"
            ],
            [
             0.9,
             "#4d9221"
            ],
            [
             1,
             "#276419"
            ]
           ],
           "sequential": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ],
           "sequentialminus": [
            [
             0,
             "#0d0887"
            ],
            [
             0.1111111111111111,
             "#46039f"
            ],
            [
             0.2222222222222222,
             "#7201a8"
            ],
            [
             0.3333333333333333,
             "#9c179e"
            ],
            [
             0.4444444444444444,
             "#bd3786"
            ],
            [
             0.5555555555555556,
             "#d8576b"
            ],
            [
             0.6666666666666666,
             "#ed7953"
            ],
            [
             0.7777777777777778,
             "#fb9f3a"
            ],
            [
             0.8888888888888888,
             "#fdca26"
            ],
            [
             1,
             "#f0f921"
            ]
           ]
          },
          "colorway": [
           "#636efa",
           "#EF553B",
           "#00cc96",
           "#ab63fa",
           "#FFA15A",
           "#19d3f3",
           "#FF6692",
           "#B6E880",
           "#FF97FF",
           "#FECB52"
          ],
          "font": {
           "color": "#2a3f5f"
          },
          "geo": {
           "bgcolor": "white",
           "lakecolor": "white",
           "landcolor": "white",
           "showlakes": true,
           "showland": true,
           "subunitcolor": "#C8D4E3"
          },
          "hoverlabel": {
           "align": "left"
          },
          "hovermode": "closest",
          "mapbox": {
           "style": "light"
          },
          "paper_bgcolor": "white",
          "plot_bgcolor": "white",
          "polar": {
           "angularaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           },
           "bgcolor": "white",
           "radialaxis": {
            "gridcolor": "#EBF0F8",
            "linecolor": "#EBF0F8",
            "ticks": ""
           }
          },
          "scene": {
           "xaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "yaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           },
           "zaxis": {
            "backgroundcolor": "white",
            "gridcolor": "#DFE8F3",
            "gridwidth": 2,
            "linecolor": "#EBF0F8",
            "showbackground": true,
            "ticks": "",
            "zerolinecolor": "#EBF0F8"
           }
          },
          "shapedefaults": {
           "line": {
            "color": "#2a3f5f"
           }
          },
          "ternary": {
           "aaxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "baxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           },
           "bgcolor": "white",
           "caxis": {
            "gridcolor": "#DFE8F3",
            "linecolor": "#A2B1C6",
            "ticks": ""
           }
          },
          "title": {
           "x": 0.05
          },
          "xaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          },
          "yaxis": {
           "automargin": true,
           "gridcolor": "#EBF0F8",
           "linecolor": "#EBF0F8",
           "ticks": "",
           "title": {
            "standoff": 15
           },
           "zerolinecolor": "#EBF0F8",
           "zerolinewidth": 2
          }
         }
        },
        "title": {
         "text": ""
        },
        "xaxis": {
         "title": {
          "text": "Hotel"
         }
        },
        "yaxis": {
         "nticks": 5,
         "title": {
          "text": "rhat_scores"
         }
        }
       }
      },
      "text/html": [
       "<div>\n",
       "        \n",
       "        \n",
       "            <div id=\"0ca56f5e-cb94-4619-a6ea-0962f9b5d211\" class=\"plotly-graph-div\" style=\"height:600px; width:100%;\"></div>\n",
       "            <script type=\"text/javascript\">\n",
       "                require([\"plotly\"], function(Plotly) {\n",
       "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
       "                    \n",
       "                if (document.getElementById(\"0ca56f5e-cb94-4619-a6ea-0962f9b5d211\")) {\n",
       "                    Plotly.newPlot(\n",
       "                        '0ca56f5e-cb94-4619-a6ea-0962f9b5d211',\n",
       "                        [{\"error_y\": {\"array\": [0.014036470388890544, 0.060212925815169425, 0.10332698899204755, 0.0959358085637333, 0.07321121520045266], \"type\": \"data\"}, \"name\": \"Desktop\", \"type\": \"bar\", \"x\": [\"ID:371\", \"ID:6085\", \"ID:367\", \"ID:1628\", \"ID:253\"], \"y\": [0.06323620701829592, 0.22016996845151438, 0.142281240497071, 0.14953048378229142, 0.3343912796236766]}, {\"error_y\": {\"array\": [0.11677800265583932, 0.055632390434544036, 0.029213313471479974, 0.07462909997845024, 0.08317211170852465], \"type\": \"data\"}, \"name\": \"Mobile\", \"type\": \"bar\", \"x\": [\"ID:371\", \"ID:6085\", \"ID:367\", \"ID:1628\", \"ID:253\"], \"y\": [0.2983235341355043, 0.24468733149546165, 0.0761359004182336, 0.15907974237842218, 0.3029204527071367]}, {\"error_y\": {\"array\": [0.18454737007101824, 0.19135757350732704, 0.17778834011917766, 0.030325356749184643, 0.1608608491907772], \"type\": \"data\"}, \"name\": \"Tablet\", \"type\": \"bar\", \"x\": [\"ID:371\", \"ID:6085\", \"ID:367\", \"ID:1628\", \"ID:253\"], \"y\": [0.17655574902892113, 0.33074622314709884, 0.13608611188828945, 0.09330697770097426, 0.34038441663142294]}],\n",
       "                        {\"font\": {\"family\": \"Courier New, monospace\", \"size\": 21}, \"height\": 600, \"legend\": {\"orientation\": \"h\", \"y\": -0.2}, \"shapes\": [{\"line\": {\"dash\": \"dot\", \"width\": 1}, \"type\": \"line\", \"x0\": 0, \"x1\": 1, \"xref\": \"paper\", \"y0\": 0.20452237459362096, \"y1\": 0.20452237459362096, \"yref\": \"y\"}], \"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"white\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"#C8D4E3\", \"linecolor\": \"#C8D4E3\", \"minorgridcolor\": \"#C8D4E3\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"white\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"#C8D4E3\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"white\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"radialaxis\": {\"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"yaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}, \"zaxis\": {\"backgroundcolor\": \"white\", \"gridcolor\": \"#DFE8F3\", \"gridwidth\": 2, \"linecolor\": \"#EBF0F8\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"#EBF0F8\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}, \"bgcolor\": \"white\", \"caxis\": {\"gridcolor\": \"#DFE8F3\", \"linecolor\": \"#A2B1C6\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"#EBF0F8\", \"linecolor\": \"#EBF0F8\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"#EBF0F8\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"\"}, \"xaxis\": {\"title\": {\"text\": \"Hotel\"}}, \"yaxis\": {\"nticks\": 5, \"title\": {\"text\": \"rhat_scores\"}}},\n",
       "                        {\"responsive\": true}\n",
       "                    ).then(function(){\n",
       "                            \n",
       "var gd = document.getElementById('0ca56f5e-cb94-4619-a6ea-0962f9b5d211');\n",
       "var x = new MutationObserver(function (mutations, observer) {{\n",
       "        var display = window.getComputedStyle(gd).display;\n",
       "        if (!display || display === 'none') {{\n",
       "            console.log([gd, 'removed!']);\n",
       "            Plotly.purge(gd);\n",
       "            observer.disconnect();\n",
       "        }}\n",
       "}});\n",
       "\n",
       "// Listen for the removal of the full notebook cells\n",
       "var notebookContainer = gd.closest('#notebook-container');\n",
       "if (notebookContainer) {{\n",
       "    x.observe(notebookContainer, {childList: true});\n",
       "}}\n",
       "\n",
       "// Listen for the clearing of the current output cell\n",
       "var outputEl = gd.closest('.output');\n",
       "if (outputEl) {{\n",
       "    x.observe(outputEl, {childList: true});\n",
       "}}\n",
       "\n",
       "                        })\n",
       "                };\n",
       "                });\n",
       "            </script>\n",
       "        </div>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plot_fairness_treatment(df_instances, input_features, input_items, top=input_items_top, title=\"\", legend={\"0\": \"Desktop\", \"1\": \"Mobile\", \"2\": \"Tablet\"})\n",
    "fig.update_layout(font={'family': 'Courier New, monospace', 'size': 21}, height=600)\n",
    "\n",
    "\n",
    "\n",
    "# Set custom x-axis labels\n",
    "fig.update_yaxes(nticks=5)\n",
    "fig.update_layout(xaxis_title=\"Hotel\")\n",
    "\n",
    "fig.show()\n",
    "#fig.show(renderer=\"svg\")\n",
    "#fig.write_image(\"output/tmp/metrics/paper/recsys_fairnes_impact_{}.eps\".format(input_features))    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.7.3 64-bit ('base': conda)",
   "language": "python",
   "name": "python37364bitbaseconda71fa0390d2e947479e80d6476a063090"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}