DataMining_Practical_Recommender_Answer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Mining - Practical Session\n",
    "\n",
    "In this hands-on session you will be taken through the process of creating movie recommendations based on the Movielens 1M dataset. \n",
    "\n",
    "To this end, we will undergo the following steps:\n",
    "1. Load the data\n",
    "2. Explore the data\n",
    "3. Pre-process the data\n",
    "4. Create a recommender model\n",
    "5. Recommend movies\n",
    "\n",
    "Let's get to it! "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First we load the packages\n",
    "Make sure that all the packages are installed correctly\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install surprise\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import defaultdict\n",
    "from sklearn import model_selection\n",
    "from sklearn.manifold import TSNE\n",
    "from surprise.model_selection import train_test_split\n",
    "from surprise.model_selection import cross_validate\n",
    "from surprise import SVD\n",
    "from surprise import Reader\n",
    "from surprise import Dataset\n",
    "from surprise.model_selection import GridSearchCV\n",
    "from surprise import PredictionImpossible\n",
    "from surprise.accuracy import rmse\n",
    "from surprise import accuracy\n",
    "from surprise import KNNWithMeans"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Load the data \n",
    "We use the Movielens 100k observations dataset which is frequently used in the recommender field. In case your computer can handle this easily, you can also work with the 1M dataset which can be found in the bu folder.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964982703</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964981247</td>\n",
       "      <td>Grumpier Old Men (1995)</td>\n",
       "      <td>Comedy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964982224</td>\n",
       "      <td>Heat (1995)</td>\n",
       "      <td>Action|Crime|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>5.0</td>\n",
       "      <td>964983815</td>\n",
       "      <td>Seven (a.k.a. Se7en) (1995)</td>\n",
       "      <td>Mystery|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>50</td>\n",
       "      <td>5.0</td>\n",
       "      <td>964982931</td>\n",
       "      <td>Usual Suspects, The (1995)</td>\n",
       "      <td>Crime|Mystery|Thriller</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   userId  movieId  rating  timestamp                        title  \\\n",
       "0       1        1     4.0  964982703             Toy Story (1995)   \n",
       "1       1        3     4.0  964981247      Grumpier Old Men (1995)   \n",
       "2       1        6     4.0  964982224                  Heat (1995)   \n",
       "3       1       47     5.0  964983815  Seven (a.k.a. Se7en) (1995)   \n",
       "4       1       50     5.0  964982931   Usual Suspects, The (1995)   \n",
       "\n",
       "                                        genres  \n",
       "0  Adventure|Animation|Children|Comedy|Fantasy  \n",
       "1                               Comedy|Romance  \n",
       "2                        Action|Crime|Thriller  \n",
       "3                             Mystery|Thriller  \n",
       "4                       Crime|Mystery|Thriller  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the data \n",
    "ratings = pd.read_csv(\"data/ratings.csv\")\n",
    "movies = pd.read_csv(\"data/movies.csv\")\n",
    "\n",
    "# add movie information to the ratings \n",
    "data = pd.merge(ratings, movies, how=\"left\")\n",
    "\n",
    "data.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That is all the data we are going to use!\n",
    "### 2. Explore the data \n",
    "Let's see how our data looks like\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are in total 100836 ratings from 9724  movies and 610 users.\n"
     ]
    }
   ],
   "source": [
    "print(\"There are in total\", len(data), \"ratings from\", data.movieId.nunique(), \" movies and\", data.userId.nunique(), \"users.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "userId         int64\n",
       "movieId        int64\n",
       "rating       float64\n",
       "timestamp      int64\n",
       "title         object\n",
       "genres        object\n",
       "dtype: object"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check if the formatting of the variables is ok\n",
    "data.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>userId</th>\n",
       "      <th>n_ratings</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>216</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   userId  n_ratings\n",
       "0       1        232\n",
       "1       2         29\n",
       "2       3         39\n",
       "3       4        216\n",
       "4       5         44"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# explore how many ratings we have per user \n",
    "user_freq = data[['userId', 'movieId']].groupby(['userId']).count().reset_index()\n",
    "user_freq.columns = ['userId', 'n_ratings']\n",
    "user_freq.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean number of ratings for a given user: 165.30.\n"
     ]
    }
   ],
   "source": [
    "print(f\"Mean number of ratings for a given user: {user_freq['n_ratings'].mean():.2f}.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_style(\"whitegrid\")\n",
    "plt.figure(figsize=(14,5))\n",
    "plt.subplot(1,2,1)\n",
    "ax = sns.countplot(x=\"rating\", data=data, palette=\"viridis\")\n",
    "plt.title(\"Distribution of movie ratings\")\n",
    "\n",
    "plt.subplot(1,2,2)\n",
    "ax = sns.kdeplot(user_freq['n_ratings'], shade=True, legend=False)\n",
    "plt.axvline(user_freq['n_ratings'].mean(), color=\"k\", linestyle=\"--\")\n",
    "plt.xlabel(\"# ratings per user\")\n",
    "plt.ylabel(\"density\")\n",
    "plt.title(\"Number of movies rated per user\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Now, explore the different features of the dataset in isolation.** You can thing of the following questions:\n",
    "1. How does the distribution look like of number of ratings per movie?\n",
    "2. Which movies have been rated most often? \n",
    "3. Which movies have the highest and lowest rating?"
   ]
  },
  {
   "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>movieId</th>\n",
       "      <th>title</th>\n",
       "      <th>n_ratings</th>\n",
       "      <th>mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>314</th>\n",
       "      <td>356</td>\n",
       "      <td>Forrest Gump (1994)</td>\n",
       "      <td>329</td>\n",
       "      <td>4.164134</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>277</th>\n",
       "      <td>318</td>\n",
       "      <td>Shawshank Redemption, The (1994)</td>\n",
       "      <td>317</td>\n",
       "      <td>4.429022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>257</th>\n",
       "      <td>296</td>\n",
       "      <td>Pulp Fiction (1994)</td>\n",
       "      <td>307</td>\n",
       "      <td>4.197068</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>510</th>\n",
       "      <td>593</td>\n",
       "      <td>Silence of the Lambs, The (1991)</td>\n",
       "      <td>279</td>\n",
       "      <td>4.161290</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1938</th>\n",
       "      <td>2571</td>\n",
       "      <td>Matrix, The (1999)</td>\n",
       "      <td>278</td>\n",
       "      <td>4.192446</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224</th>\n",
       "      <td>260</td>\n",
       "      <td>Star Wars: Episode IV - A New Hope (1977)</td>\n",
       "      <td>251</td>\n",
       "      <td>4.231076</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>418</th>\n",
       "      <td>480</td>\n",
       "      <td>Jurassic Park (1993)</td>\n",
       "      <td>238</td>\n",
       "      <td>3.750000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>110</td>\n",
       "      <td>Braveheart (1995)</td>\n",
       "      <td>237</td>\n",
       "      <td>4.031646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>507</th>\n",
       "      <td>589</td>\n",
       "      <td>Terminator 2: Judgment Day (1991)</td>\n",
       "      <td>224</td>\n",
       "      <td>3.970982</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>461</th>\n",
       "      <td>527</td>\n",
       "      <td>Schindler's List (1993)</td>\n",
       "      <td>220</td>\n",
       "      <td>4.225000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      movieId                                      title  n_ratings      mean\n",
       "314       356                        Forrest Gump (1994)        329  4.164134\n",
       "277       318           Shawshank Redemption, The (1994)        317  4.429022\n",
       "257       296                        Pulp Fiction (1994)        307  4.197068\n",
       "510       593           Silence of the Lambs, The (1991)        279  4.161290\n",
       "1938     2571                         Matrix, The (1999)        278  4.192446\n",
       "224       260  Star Wars: Episode IV - A New Hope (1977)        251  4.231076\n",
       "418       480                       Jurassic Park (1993)        238  3.750000\n",
       "97        110                          Braveheart (1995)        237  4.031646\n",
       "507       589          Terminator 2: Judgment Day (1991)        224  3.970982\n",
       "461       527                    Schindler's List (1993)        220  4.225000"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# explore how many ratings we have per movie and most rated movies \n",
    "movie_freq = data[['userId', 'movieId', 'title','rating']].groupby(['movieId','title']).agg([\"count\",\"mean\"]).reset_index()\n",
    "movie_freq.columns = movie_freq.columns.map('|'.join).str.strip('|')\n",
    "movie_freq = movie_freq.drop([\"userId|mean\",\"rating|count\"], axis=1)\n",
    "movie_freq.columns = ['movieId','title','n_ratings','mean']\n",
    "movie_freq.sort_values(\"n_ratings\", ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average number of ratings per movie: 10.369806663924312\n"
     ]
    }
   ],
   "source": [
    "# distribution of number of ratings per movie\n",
    "ax = sns.kdeplot(movie_freq['n_ratings'], shade=True, legend=False)\n",
    "plt.axvline(movie_freq['n_ratings'].mean(), color=\"k\", linestyle=\"--\")\n",
    "plt.xlabel(\"# ratings per movie\")\n",
    "plt.ylabel(\"density\")\n",
    "plt.title(\"Number of ratings per movie\")\n",
    "plt.show()\n",
    "\n",
    "print(\"Average number of ratings per movie:\", movie_freq.n_ratings.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>movieId</th>\n",
       "      <th>title</th>\n",
       "      <th>n_ratings</th>\n",
       "      <th>mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>277</th>\n",
       "      <td>318</td>\n",
       "      <td>Shawshank Redemption, The (1994)</td>\n",
       "      <td>317</td>\n",
       "      <td>4.429022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>659</th>\n",
       "      <td>858</td>\n",
       "      <td>Godfather, The (1972)</td>\n",
       "      <td>192</td>\n",
       "      <td>4.289062</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2224</th>\n",
       "      <td>2959</td>\n",
       "      <td>Fight Club (1999)</td>\n",
       "      <td>218</td>\n",
       "      <td>4.272936</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>974</th>\n",
       "      <td>1276</td>\n",
       "      <td>Cool Hand Luke (1967)</td>\n",
       "      <td>57</td>\n",
       "      <td>4.271930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>602</th>\n",
       "      <td>750</td>\n",
       "      <td>Dr. Strangelove or: How I Learned to Stop Worr...</td>\n",
       "      <td>97</td>\n",
       "      <td>4.268041</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>686</th>\n",
       "      <td>904</td>\n",
       "      <td>Rear Window (1954)</td>\n",
       "      <td>84</td>\n",
       "      <td>4.261905</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>921</th>\n",
       "      <td>1221</td>\n",
       "      <td>Godfather: Part II, The (1974)</td>\n",
       "      <td>129</td>\n",
       "      <td>4.259690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6298</th>\n",
       "      <td>48516</td>\n",
       "      <td>Departed, The (2006)</td>\n",
       "      <td>107</td>\n",
       "      <td>4.252336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>913</th>\n",
       "      <td>1213</td>\n",
       "      <td>Goodfellas (1990)</td>\n",
       "      <td>126</td>\n",
       "      <td>4.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>694</th>\n",
       "      <td>912</td>\n",
       "      <td>Casablanca (1942)</td>\n",
       "      <td>100</td>\n",
       "      <td>4.240000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      movieId                                              title  n_ratings  \\\n",
       "277       318                   Shawshank Redemption, The (1994)        317   \n",
       "659       858                              Godfather, The (1972)        192   \n",
       "2224     2959                                  Fight Club (1999)        218   \n",
       "974      1276                              Cool Hand Luke (1967)         57   \n",
       "602       750  Dr. Strangelove or: How I Learned to Stop Worr...         97   \n",
       "686       904                                 Rear Window (1954)         84   \n",
       "921      1221                     Godfather: Part II, The (1974)        129   \n",
       "6298    48516                               Departed, The (2006)        107   \n",
       "913      1213                                  Goodfellas (1990)        126   \n",
       "694       912                                  Casablanca (1942)        100   \n",
       "\n",
       "          mean  \n",
       "277   4.429022  \n",
       "659   4.289062  \n",
       "2224  4.272936  \n",
       "974   4.271930  \n",
       "602   4.268041  \n",
       "686   4.261905  \n",
       "921   4.259690  \n",
       "6298  4.252336  \n",
       "913   4.250000  \n",
       "694   4.240000  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# best rated movies with minimal 50 ratings\n",
    "movie_freq[movie_freq[\"n_ratings\"]>50].sort_values(\"mean\", ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>movieId</th>\n",
       "      <th>title</th>\n",
       "      <th>n_ratings</th>\n",
       "      <th>mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2028</th>\n",
       "      <td>2701</td>\n",
       "      <td>Wild Wild West (1999)</td>\n",
       "      <td>53</td>\n",
       "      <td>2.207547</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>379</th>\n",
       "      <td>435</td>\n",
       "      <td>Coneheads (1993)</td>\n",
       "      <td>63</td>\n",
       "      <td>2.420635</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>376</th>\n",
       "      <td>432</td>\n",
       "      <td>City Slickers II: The Legend of Curly's Gold (...</td>\n",
       "      <td>55</td>\n",
       "      <td>2.645455</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>145</th>\n",
       "      <td>173</td>\n",
       "      <td>Judge Dredd (1995)</td>\n",
       "      <td>62</td>\n",
       "      <td>2.669355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>172</td>\n",
       "      <td>Johnny Mnemonic (1995)</td>\n",
       "      <td>53</td>\n",
       "      <td>2.679245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>559</th>\n",
       "      <td>673</td>\n",
       "      <td>Space Jam (1996)</td>\n",
       "      <td>53</td>\n",
       "      <td>2.707547</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2696</th>\n",
       "      <td>3623</td>\n",
       "      <td>Mission: Impossible II (2000)</td>\n",
       "      <td>77</td>\n",
       "      <td>2.714286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>364</th>\n",
       "      <td>420</td>\n",
       "      <td>Beverly Hills Cop III (1994)</td>\n",
       "      <td>59</td>\n",
       "      <td>2.720339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2962</th>\n",
       "      <td>3977</td>\n",
       "      <td>Charlie's Angels (2000)</td>\n",
       "      <td>72</td>\n",
       "      <td>2.722222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>19</td>\n",
       "      <td>Ace Ventura: When Nature Calls (1995)</td>\n",
       "      <td>88</td>\n",
       "      <td>2.727273</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      movieId                                              title  n_ratings  \\\n",
       "2028     2701                              Wild Wild West (1999)         53   \n",
       "379       435                                   Coneheads (1993)         63   \n",
       "376       432  City Slickers II: The Legend of Curly's Gold (...         55   \n",
       "145       173                                 Judge Dredd (1995)         62   \n",
       "144       172                             Johnny Mnemonic (1995)         53   \n",
       "559       673                                   Space Jam (1996)         53   \n",
       "2696     3623                      Mission: Impossible II (2000)         77   \n",
       "364       420                       Beverly Hills Cop III (1994)         59   \n",
       "2962     3977                            Charlie's Angels (2000)         72   \n",
       "18         19              Ace Ventura: When Nature Calls (1995)         88   \n",
       "\n",
       "          mean  \n",
       "2028  2.207547  \n",
       "379   2.420635  \n",
       "376   2.645455  \n",
       "145   2.669355  \n",
       "144   2.679245  \n",
       "559   2.707547  \n",
       "2696  2.714286  \n",
       "364   2.720339  \n",
       "2962  2.722222  \n",
       "18    2.727273  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# worst rated movies with minimal 50 ratings\n",
    "movie_freq[movie_freq[\"n_ratings\"]>50].sort_values(\"mean\", ascending=True).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Think about what this data exploration means for the model.** Some questions:\n",
    "1. Most ratings are 4, what could then be a good random benchmark algorithm?\n",
    "2. If people only rated 3 movies, should we include them?\n",
    "3. Same for movies: when to include them?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Pre-process the data\n",
    "\n",
    "1. Filter out the movies and users you want to include in the dataset\n",
    "2. Calculate the sparsity of the data\n",
    "3. Create a train and test set\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are in total 94794 ratings from 4980  movies and 610 users.\n"
     ]
    }
   ],
   "source": [
    "min_number = 3\n",
    "\n",
    "users_more_ratings = data.groupby(\"userId\").count().reset_index().sort_values(\"movieId\")\n",
    "users_more_ratings = users_more_ratings[users_more_ratings[\"movieId\"]>=min_number].userId.unique()\n",
    "\n",
    "movies_more_ratings = data.groupby(\"movieId\").count().reset_index().sort_values(\"userId\")\n",
    "movies_more_ratings = movies_more_ratings[movies_more_ratings[\"userId\"]>=min_number].movieId.unique()\n",
    "\n",
    "data = data[data[\"movieId\"].isin(movies_more_ratings) & data[\"userId\"].isin(users_more_ratings)]\n",
    "print(\"There are in total\", len(data), \"ratings from\", data.movieId.nunique(), \" movies and\", data.userId.nunique(), \"users.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sparsity:  0.031204819277108435\n"
     ]
    }
   ],
   "source": [
    "# Calculate Sparsity\n",
    "print(\"Sparsity: \", len(data)/ (data.movieId.nunique() *data.userId.nunique()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Create a recommender model\n",
    "**For this hands on we use the library \"surprise\" to create our recommender models** \n",
    "For more information check out this webste: https://surprise.readthedocs.io/en/stable/\n",
    "\n",
    "We also restrict ourselves to collaborative filtering methods only. \n",
    "Remember, that means that we don't take user and movie features into account but only the matrix with ratings. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This is how we load the data. \n",
    "\n",
    "Matrix = data[[\"userId\",\"movieId\",\"rating\"]]\n",
    "reader = Reader(rating_scale=(1,5))\n",
    "matrix = Dataset.load_from_df(Matrix, reader)\n",
    "\n",
    "# Then we create a 80-20 split of train and test data\n",
    "train_df, test_df = train_test_split(matrix, test_size=0.2, random_state=100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1 Collaborative filtering: memory based - user\n",
    "We use a K-NN method where we subtract the mean rating of every user. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing the cosine similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "RMSE: 0.7130\n",
      "RMSE on train data 0.7129830710070995\n",
      "RMSE: 0.8892\n",
      "RMSE on test data 0.8891940172573102\n"
     ]
    }
   ],
   "source": [
    "sim_options = {'name': 'cosine', 'min_support':1 ,'user_based': True }\n",
    "recommender_KNN = KNNWithMeans(k=10, min_k=1, sim_options=sim_options, verbose=True)\n",
    "recommender_KNN.fit(train_df)\n",
    "\n",
    "original_train = train_df.build_testset()\n",
    "predictions_train = recommender_KNN.test(original_train)\n",
    "predictions = recommender_KNN.test(test_df)\n",
    "\n",
    "print(\"RMSE on train data\", accuracy.rmse(predictions_train, verbose=True))\n",
    "print(\"RMSE on test data\",accuracy.rmse(predictions, verbose=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**As you can see, there are a lot of parameters that can affect the algorithm performance and explain the overfitting on the test set. Try out some parameter values and see which one has a lot of effect**\n",
    "If you change the min_k or the min_support, it might happen that some recommendations are not possible to give. The min_support for example tells how many items two users need to have in common before a similarity can be calculated. If you make recommendations with such a model, make sure to filter them out in the end (in the details column on the prediction) \n",
    "\n",
    "**Let's do a parameter sweep and investigate the best parameters**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "Computing the msd similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "0.8835457114550226\n",
      "{'name': 'pearson', 'min_support': 1, 'user_based': 'True', 'k': 60, 'min_k': 1}\n"
     ]
    }
   ],
   "source": [
    "param_grid = {'name': ['pearson', 'cosine'], 'min_support': [1], \"user_based\": [\"True\", \"False\"]\n",
    "              , 'k':[10,40,60], 'min_k':[1]}\n",
    "gs = GridSearchCV(KNNWithMeans, param_grid, measures=['rmse'], cv=3)\n",
    "gs.fit(matrix)\n",
    "\n",
    "# best RMSE score\n",
    "print(gs.best_score['rmse'])\n",
    "# combination of parameters that gave the best RMSE score\n",
    "print(gs.best_params['rmse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing the pearson similarity matrix...\n",
      "Done computing similarity matrix.\n",
      "RMSE: 0.6115\n",
      "RMSE on train data 0.6114689160715127\n",
      "RMSE: 0.8747\n",
      "RMSE on test data 0.874744528324542\n"
     ]
    }
   ],
   "source": [
    "# Use these best parameters on the previous set up with a train and test split\n",
    "sim_options = {'name': 'pearson', 'min_support':1 ,'user_based': True }\n",
    "recommender_KNN = KNNWithMeans(k=60, min_k=1, sim_options=sim_options, verbose=True)\n",
    "recommender_KNN.fit(train_df)\n",
    "\n",
    "original_train = train_df.build_testset()\n",
    "predictions_train = recommender_KNN.test(original_train)\n",
    "predictions = recommender_KNN.test(test_df)\n",
    "\n",
    "print(\"RMSE on train data\", accuracy.rmse(predictions_train, verbose=True))\n",
    "print(\"RMSE on test data\",accuracy.rmse(predictions, verbose=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### This is indeed a bit better! \n",
    "Let's investigate some more:\n",
    "1. How does the confusion matrices look like (suppose that a rating of 3.5 or higher was indeed a good recommendation). \n",
    "2. How do we do compared to a benchmark that always predicts 3.5 (in terms of precision, recall and accuracy on test set)?\n",
    "3. What are the best and worst predictions in the train and test set?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# First prep the data\n",
    "predictions_train = pd.DataFrame(predictions_train, columns=['userId', 'movieId', 'rating', 'rating_p', 'details'])\n",
    "predictions_test = pd.DataFrame(predictions, columns=['userId', 'movieId', 'rating', 'rating_p', 'details'])\n",
    "train_matrix = pd.DataFrame(original_train, columns=['userId', 'movieId', 'rating'])\n",
    "test_matrix = pd.DataFrame(test_df, columns=['userId', 'movieId', 'rating'])\n",
    "\n",
    "train_matrix = pd.merge(train_matrix,predictions_train, on= ['userId', 'movieId', 'rating'], how=\"left\")\n",
    "test_matrix = pd.merge(test_matrix,predictions_test, on= ['userId', 'movieId', 'rating'], how=\"left\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train\n",
      "Predicted    0.0    1.0\n",
      "Actual                 \n",
      "0.0        24869   4165\n",
      "1.0         9262  37539\n",
      "Accuracy 0.82\n",
      "Recall 0.8\n",
      "Precision 0.9\n",
      "\n",
      "---Test\n",
      "Predicted   0.0   1.0\n",
      "Actual               \n",
      "0.0        5113  2130\n",
      "1.0        3230  8486\n",
      "Accuracy 0.72\n",
      "Recall 0.72\n",
      "Precision 0.8\n"
     ]
    }
   ],
   "source": [
    "# Create threshold values\n",
    "train_matrix.loc[train_matrix[\"rating\"] >=3.5,'rating_th'] = 1\n",
    "train_matrix.loc[train_matrix[\"rating\"] <3.5,'rating_th'] = 0\n",
    "train_matrix.loc[train_matrix[\"rating_p\"] >=3.5,'rating_p_th'] = 1\n",
    "train_matrix.loc[train_matrix[\"rating_p\"] <3.5,'rating_p_th'] = 0\n",
    "\n",
    "test_matrix.loc[test_matrix[\"rating\"] >=3.5,'rating_th'] = 1\n",
    "test_matrix.loc[test_matrix[\"rating\"] <3.5,'rating_th'] = 0\n",
    "test_matrix.loc[test_matrix[\"rating_p\"] >=3.5,'rating_p_th'] = 1\n",
    "test_matrix.loc[test_matrix[\"rating_p\"] <3.5,'rating_p_th'] = 0\n",
    "test_matrix['benchmark_35'] = 1\n",
    "\n",
    "confusion_matrix_train = pd.crosstab(train_matrix['rating_th'], train_matrix['rating_p_th'], rownames=['Actual'], colnames=['Predicted'])\n",
    "confusion_matrix_test = pd.crosstab(test_matrix['rating_th'], test_matrix['rating_p_th'], rownames=['Actual'], colnames=['Predicted'])\n",
    "\n",
    "\n",
    "print(\"train\")\n",
    "print(confusion_matrix_train)\n",
    "print(\"Accuracy\", np.round((confusion_matrix_train.loc[0,0] + confusion_matrix_train.loc[1,1])/ confusion_matrix_train.sum().sum(),2)  ) \n",
    "print(\"Recall\", np.round( confusion_matrix_train.loc[1,1] /(confusion_matrix_train.loc[1,1]+confusion_matrix_train.loc[1,0]), 2))\n",
    "print(\"Precision\", np.round( confusion_matrix_train.loc[1,1] /(confusion_matrix_train.loc[1,1]+confusion_matrix_train.loc[0,1]), 2))\n",
    "\n",
    "print(\"\\n---Test\")\n",
    "print(confusion_matrix_test)\n",
    "print(\"Accuracy\", np.round((confusion_matrix_test.loc[0,0] + confusion_matrix_test.loc[1,1])/ confusion_matrix_test.sum().sum(),2)  ) \n",
    "print(\"Recall\", np.round( confusion_matrix_test.loc[1,1] /(confusion_matrix_test.loc[1,1]+confusion_matrix_test.loc[1,0]), 2))\n",
    "print(\"Precision\", np.round( confusion_matrix_test.loc[1,1] /(confusion_matrix_test.loc[1,1]+confusion_matrix_test.loc[0,1]), 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted      1\n",
      "Actual          \n",
      "0.0         7243\n",
      "1.0        11716\n",
      "Accuracy/Precision: 0.62\n",
      "Recall: 1\n"
     ]
    }
   ],
   "source": [
    "confusion_matrix_test_35 = pd.crosstab(test_matrix['rating_th'], test_matrix['benchmark_35'], rownames=['Actual'], colnames=['Predicted'])\n",
    "print(confusion_matrix_test_35)\n",
    "print(\"Accuracy/Precision:\", np.round(confusion_matrix_test_35.loc[1,1]/ confusion_matrix_test_35.sum().sum(),2)  ) \n",
    "print(\"Recall:\", 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>rating_p</th>\n",
       "      <th>details</th>\n",
       "      <th>rating_th</th>\n",
       "      <th>rating_p_th</th>\n",
       "      <th>error</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>52726</th>\n",
       "      <td>573</td>\n",
       "      <td>3996</td>\n",
       "      <td>0.5</td>\n",
       "      <td>3.900735</td>\n",
       "      <td>{'actual_k': 60, 'was_impossible': False}</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.400735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52614</th>\n",
       "      <td>573</td>\n",
       "      <td>44199</td>\n",
       "      <td>0.5</td>\n",
       "      <td>3.891924</td>\n",
       "      <td>{'actual_k': 29, 'was_impossible': False}</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.391924</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3506</th>\n",
       "      <td>393</td>\n",
       "      <td>778</td>\n",
       "      <td>0.5</td>\n",
       "      <td>3.791024</td>\n",
       "      <td>{'actual_k': 57, 'was_impossible': False}</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.291024</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       userId  movieId  rating  rating_p  \\\n",
       "52726     573     3996     0.5  3.900735   \n",
       "52614     573    44199     0.5  3.891924   \n",
       "3506      393      778     0.5  3.791024   \n",
       "\n",
       "                                         details  rating_th  rating_p_th  \\\n",
       "52726  {'actual_k': 60, 'was_impossible': False}        0.0          1.0   \n",
       "52614  {'actual_k': 29, 'was_impossible': False}        0.0          1.0   \n",
       "3506   {'actual_k': 57, 'was_impossible': False}        0.0          1.0   \n",
       "\n",
       "          error  \n",
       "52726  3.400735  \n",
       "52614  3.391924  \n",
       "3506   3.291024  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Let's look at some of the biggest errors\n",
    "train_matrix[\"error\"] = np.abs(train_matrix[\"rating_p\"] - train_matrix[\"rating\"])\n",
    "train_matrix.sort_values(\"error\", ascending=False).head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "movie\n",
      "Ratings of user:\n",
      "count    247.000000\n",
      "mean       4.157895\n",
      "std        1.020804\n",
      "min        0.500000\n",
      "25%        4.000000\n",
      "50%        4.500000\n",
      "75%        5.000000\n",
      "max        5.000000\n",
      "Name: rating, dtype: float64\n",
      "Ratings of movie:\n",
      "count    95.000000\n",
      "mean      3.805263\n",
      "std       1.062651\n",
      "min       0.500000\n",
      "25%       3.500000\n",
      "50%       4.000000\n",
      "75%       4.500000\n",
      "max       5.000000\n",
      "Name: rating, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# Let's check why the first prediction is so off\n",
    "print(\"movie\")\n",
    "data.loc[(data[\"movieId\"]==3996),\"title\"].head(1)\n",
    "print(\"Ratings of user:\")\n",
    "print( train_matrix[(train_matrix[\"userId\"]==573)].rating.describe())\n",
    "print(\"Ratings of movie:\")\n",
    "print(train_matrix[(train_matrix[\"movieId\"]==3996)].rating.describe())\n",
    "# Conclusion: the user gives good rates in general and the movie, although not much ratings, has been rated well too."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2 Collaborative filtering: model based - MF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE: 0.7302\n",
      "RMSE on train data 0.7302286935626725\n",
      "RMSE: 0.8786\n",
      "RMSE on test data 0.8785901982054994\n"
     ]
    }
   ],
   "source": [
    "# We'll use the SVD algorithm that uses SGD to learn the weights\n",
    "recommender_SVD = SVD(n_factors= 5, n_epochs = 10, reg_all = 0.1, lr_all = 0.1, verbose=False, biased=True)\n",
    "recommender_SVD.fit(train_df)\n",
    "\n",
    "original_train = train_df.build_testset()\n",
    "predictions_train_SVD = recommender_SVD.test(original_train)\n",
    "predictions_SVD = recommender_SVD.test(test_df)\n",
    "\n",
    "print(\"RMSE on train data\", accuracy.rmse(predictions_train_SVD, verbose=True))\n",
    "print(\"RMSE on test data\",accuracy.rmse(predictions_SVD, verbose=True))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Repeat the exercise from above for this method**\n",
    "Does this method give better results?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8777483675848017\n",
      "{'n_factors': 5, 'n_epochs': 10, 'reg_all': 0.05, 'lr_all': 0.05}\n"
     ]
    }
   ],
   "source": [
    "param_grid = {'n_factors': [5, 10,30], 'n_epochs': [10,30], \"reg_all\": [0.05, 0.02]\n",
    "              , 'lr_all':[0.1,0.05]}\n",
    "gs = GridSearchCV(SVD, param_grid, measures=['rmse'], cv=3)\n",
    "gs.fit(matrix)\n",
    "\n",
    "# best RMSE score\n",
    "print(gs.best_score['rmse'])\n",
    "# combination of parameters that gave the best RMSE score\n",
    "print(gs.best_params['rmse'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Recommend movies\n",
    "\n",
    "Let's take 1 user and use both algorithms to predict the recommended movies for these users"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "user raw id: 41 item raw id:  3766\n",
      "KNN:  user: 555        item: 7121       r_ui = None   est = 4.54   {'actual_k': 3, 'was_impossible': False}\n",
      "SVD:  user: 555        item: 7121       r_ui = None   est = 4.46   {'was_impossible': False}\n"
     ]
    }
   ],
   "source": [
    "# Make prediction and are they different?\n",
    "u = train_df.to_inner_uid(555)\n",
    "i = train_df.to_inner_iid(7121)\n",
    "print(\"user raw id:\" , u, \"item raw id: \",i)\n",
    "user = train_df.to_raw_uid(u)\n",
    "item = train_df.to_raw_iid(i)\n",
    "\n",
    "# going to inner and back to raw id seems a bit redundant but using the raw id's immediately is not regocnised\n",
    "print(\"KNN: \", recommender_KNN.predict(user, item))  \n",
    "print(\"SVD: \", recommender_SVD.predict(user, item))  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Let's look at the top recommendations for user 555\n",
    "make_recommendations = train_df.build_anti_testset()\n",
    "recommendations = recommender_SVD.test(make_recommendations)\n",
    "\n",
    "top_n = defaultdict(list)\n",
    "for uid, iid, true_r, est, _ in recommendations:\n",
    "    top_n[uid].append((iid, est))\n",
    "    \n",
    "# Then sort the predictions for each user and retrieve the k highest ones.\n",
    "for uid, user_ratings in top_n.items():\n",
    "    user_ratings.sort(key=lambda x: x[1], reverse=True)\n",
    "    top_n[uid] = user_ratings[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(26810, 4.9694726633171245),\n",
       " (51931, 4.826745512780749),\n",
       " (1172, 4.820813871928653),\n",
       " (174053, 4.760040343207083),\n",
       " (1237, 4.757206124051348),\n",
       " (2239, 4.723501273918199),\n",
       " (177593, 4.722113558950766),\n",
       " (2732, 4.709352876908652),\n",
       " (86781, 4.684823739004687),\n",
       " (6300, 4.659407321239786)]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_n[555]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**We can also explore the model a bit and look at similar users**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "user id:  439\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1, 80, 86, 103, 127, 146, 183, 224, 235, 236]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inner_user_id = recommender_KNN.trainset.to_inner_uid(487)\n",
    "print(\"user id: \", inner_user_id)\n",
    "user_neigh = recommender_KNN.get_neighbors(inner_user_id, k=10)\n",
    "user_neigh"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>85983</th>\n",
       "      <td>558</td>\n",
       "      <td>5420</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035415871</td>\n",
       "      <td>Windtalkers (2002)</td>\n",
       "      <td>Action|Drama|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85966</th>\n",
       "      <td>558</td>\n",
       "      <td>5064</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035415811</td>\n",
       "      <td>The Count of Monte Cristo (2002)</td>\n",
       "      <td>Action|Adventure|Drama|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85957</th>\n",
       "      <td>558</td>\n",
       "      <td>3831</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035415531</td>\n",
       "      <td>Saving Grace (2000)</td>\n",
       "      <td>Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85959</th>\n",
       "      <td>558</td>\n",
       "      <td>4159</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416048</td>\n",
       "      <td>3000 Miles to Graceland (2001)</td>\n",
       "      <td>Action|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85961</th>\n",
       "      <td>558</td>\n",
       "      <td>4306</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416189</td>\n",
       "      <td>Shrek (2001)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy|Ro...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85950</th>\n",
       "      <td>558</td>\n",
       "      <td>3186</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416294</td>\n",
       "      <td>Girl, Interrupted (1999)</td>\n",
       "      <td>Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85962</th>\n",
       "      <td>558</td>\n",
       "      <td>4308</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416273</td>\n",
       "      <td>Moulin Rouge (2001)</td>\n",
       "      <td>Drama|Musical|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85947</th>\n",
       "      <td>558</td>\n",
       "      <td>2568</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416048</td>\n",
       "      <td>Mod Squad, The (1999)</td>\n",
       "      <td>Action|Crime</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85946</th>\n",
       "      <td>558</td>\n",
       "      <td>2297</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035416242</td>\n",
       "      <td>What Dreams May Come (1998)</td>\n",
       "      <td>Adventure|Drama|Fantasy|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85965</th>\n",
       "      <td>558</td>\n",
       "      <td>4993</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1035415811</td>\n",
       "      <td>Lord of the Rings: The Fellowship of the Ring,...</td>\n",
       "      <td>Adventure|Fantasy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       userId  movieId  rating   timestamp  \\\n",
       "85983     558     5420     5.0  1035415871   \n",
       "85966     558     5064     5.0  1035415811   \n",
       "85957     558     3831     5.0  1035415531   \n",
       "85959     558     4159     5.0  1035416048   \n",
       "85961     558     4306     5.0  1035416189   \n",
       "85950     558     3186     5.0  1035416294   \n",
       "85962     558     4308     5.0  1035416273   \n",
       "85947     558     2568     5.0  1035416048   \n",
       "85946     558     2297     5.0  1035416242   \n",
       "85965     558     4993     5.0  1035415811   \n",
       "\n",
       "                                                   title  \\\n",
       "85983                                 Windtalkers (2002)   \n",
       "85966                   The Count of Monte Cristo (2002)   \n",
       "85957                                Saving Grace (2000)   \n",
       "85959                     3000 Miles to Graceland (2001)   \n",
       "85961                                       Shrek (2001)   \n",
       "85950                           Girl, Interrupted (1999)   \n",
       "85962                                Moulin Rouge (2001)   \n",
       "85947                              Mod Squad, The (1999)   \n",
       "85946                        What Dreams May Come (1998)   \n",
       "85965  Lord of the Rings: The Fellowship of the Ring,...   \n",
       "\n",
       "                                                  genres  \n",
       "85983                                   Action|Drama|War  \n",
       "85966                    Action|Adventure|Drama|Thriller  \n",
       "85957                                             Comedy  \n",
       "85959                                    Action|Thriller  \n",
       "85961  Adventure|Animation|Children|Comedy|Fantasy|Ro...  \n",
       "85950                                              Drama  \n",
       "85962                              Drama|Musical|Romance  \n",
       "85947                                       Action|Crime  \n",
       "85946                    Adventure|Drama|Fantasy|Romance  \n",
       "85965                                  Adventure|Fantasy  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Let's look at the movies of user 1\n",
    "user_sim = train_df.to_raw_uid(1)\n",
    "data[data[\"userId\"]==user_sim].sort_values(\"rating\", ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "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>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>78340</th>\n",
       "      <td>487</td>\n",
       "      <td>27611</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1428574594</td>\n",
       "      <td>Battlestar Galactica (2003)</td>\n",
       "      <td>Drama|Sci-Fi|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78359</th>\n",
       "      <td>487</td>\n",
       "      <td>116797</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1428650174</td>\n",
       "      <td>The Imitation Game (2014)</td>\n",
       "      <td>Drama|Thriller|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78321</th>\n",
       "      <td>487</td>\n",
       "      <td>4878</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574246</td>\n",
       "      <td>Donnie Darko (2001)</td>\n",
       "      <td>Drama|Mystery|Sci-Fi|Thriller</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78357</th>\n",
       "      <td>487</td>\n",
       "      <td>109487</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574546</td>\n",
       "      <td>Interstellar (2014)</td>\n",
       "      <td>Sci-Fi|IMAX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78349</th>\n",
       "      <td>487</td>\n",
       "      <td>81845</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574557</td>\n",
       "      <td>King's Speech, The (2010)</td>\n",
       "      <td>Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78347</th>\n",
       "      <td>487</td>\n",
       "      <td>61240</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574619</td>\n",
       "      <td>Let the Right One In (Låt den rätte komma in) ...</td>\n",
       "      <td>Drama|Fantasy|Horror|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78345</th>\n",
       "      <td>487</td>\n",
       "      <td>56367</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428650046</td>\n",
       "      <td>Juno (2007)</td>\n",
       "      <td>Comedy|Drama|Romance</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78339</th>\n",
       "      <td>487</td>\n",
       "      <td>8961</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574243</td>\n",
       "      <td>Incredibles, The (2004)</td>\n",
       "      <td>Action|Adventure|Animation|Children|Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78338</th>\n",
       "      <td>487</td>\n",
       "      <td>8874</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574269</td>\n",
       "      <td>Shaun of the Dead (2004)</td>\n",
       "      <td>Comedy|Horror</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78337</th>\n",
       "      <td>487</td>\n",
       "      <td>7438</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1428574247</td>\n",
       "      <td>Kill Bill: Vol. 2 (2004)</td>\n",
       "      <td>Action|Drama|Thriller</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       userId  movieId  rating   timestamp  \\\n",
       "78340     487    27611     5.0  1428574594   \n",
       "78359     487   116797     4.5  1428650174   \n",
       "78321     487     4878     4.0  1428574246   \n",
       "78357     487   109487     4.0  1428574546   \n",
       "78349     487    81845     4.0  1428574557   \n",
       "78347     487    61240     4.0  1428574619   \n",
       "78345     487    56367     4.0  1428650046   \n",
       "78339     487     8961     4.0  1428574243   \n",
       "78338     487     8874     4.0  1428574269   \n",
       "78337     487     7438     4.0  1428574247   \n",
       "\n",
       "                                                   title  \\\n",
       "78340                        Battlestar Galactica (2003)   \n",
       "78359                          The Imitation Game (2014)   \n",
       "78321                                Donnie Darko (2001)   \n",
       "78357                                Interstellar (2014)   \n",
       "78349                          King's Speech, The (2010)   \n",
       "78347  Let the Right One In (Låt den rätte komma in) ...   \n",
       "78345                                        Juno (2007)   \n",
       "78339                            Incredibles, The (2004)   \n",
       "78338                           Shaun of the Dead (2004)   \n",
       "78337                           Kill Bill: Vol. 2 (2004)   \n",
       "\n",
       "                                           genres  \n",
       "78340                            Drama|Sci-Fi|War  \n",
       "78359                          Drama|Thriller|War  \n",
       "78321               Drama|Mystery|Sci-Fi|Thriller  \n",
       "78357                                 Sci-Fi|IMAX  \n",
       "78349                                       Drama  \n",
       "78347                Drama|Fantasy|Horror|Romance  \n",
       "78345                        Comedy|Drama|Romance  \n",
       "78339  Action|Adventure|Animation|Children|Comedy  \n",
       "78338                               Comedy|Horror  \n",
       "78337                       Action|Drama|Thriller  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# and user 439\n",
    "sim_movie = train_df.to_raw_uid(439)\n",
    "data[data[\"userId\"]==sim_movie].sort_values(\"rating\", ascending=False).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**We can also check out which movies are similar looking at the latent vector spaces**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "latent_movie = pd.DataFrame(recommender_SVD.qi, columns=['latent_1', 'latent_2', 'latent_3','latent_4','latent_5']) # for user latent factors use recommender_SVD.pu\n",
    "latent_movie = latent_movie.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>index</th>\n",
       "      <th>latent_1</th>\n",
       "      <th>latent_2</th>\n",
       "      <th>latent_3</th>\n",
       "      <th>latent_4</th>\n",
       "      <th>latent_5</th>\n",
       "      <th>movieId</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>-0.155737</td>\n",
       "      <td>-0.074751</td>\n",
       "      <td>-0.167568</td>\n",
       "      <td>-0.016852</td>\n",
       "      <td>0.014442</td>\n",
       "      <td>4010</td>\n",
       "      <td>Brewster's Millions (1985)</td>\n",
       "      <td>Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>-0.142615</td>\n",
       "      <td>-0.003803</td>\n",
       "      <td>0.004407</td>\n",
       "      <td>-0.013462</td>\n",
       "      <td>0.386189</td>\n",
       "      <td>3285</td>\n",
       "      <td>Beach, The (2000)</td>\n",
       "      <td>Adventure|Drama</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>-0.436258</td>\n",
       "      <td>0.864932</td>\n",
       "      <td>-0.341606</td>\n",
       "      <td>-0.017162</td>\n",
       "      <td>0.112548</td>\n",
       "      <td>1272</td>\n",
       "      <td>Patton (1970)</td>\n",
       "      <td>Drama|War</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>-0.484146</td>\n",
       "      <td>-0.230409</td>\n",
       "      <td>0.225369</td>\n",
       "      <td>0.180206</td>\n",
       "      <td>-0.142160</td>\n",
       "      <td>3253</td>\n",
       "      <td>Wayne's World (1992)</td>\n",
       "      <td>Comedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>-0.089072</td>\n",
       "      <td>-0.350937</td>\n",
       "      <td>0.353257</td>\n",
       "      <td>0.366479</td>\n",
       "      <td>0.009860</td>\n",
       "      <td>1084</td>\n",
       "      <td>Bonnie and Clyde (1967)</td>\n",
       "      <td>Crime|Drama</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index  latent_1  latent_2  latent_3  latent_4  latent_5  movieId  \\\n",
       "0      0 -0.155737 -0.074751 -0.167568 -0.016852  0.014442     4010   \n",
       "1      1 -0.142615 -0.003803  0.004407 -0.013462  0.386189     3285   \n",
       "2      2 -0.436258  0.864932 -0.341606 -0.017162  0.112548     1272   \n",
       "3      3 -0.484146 -0.230409  0.225369  0.180206 -0.142160     3253   \n",
       "4      4 -0.089072 -0.350937  0.353257  0.366479  0.009860     1084   \n",
       "\n",
       "                        title           genres  \n",
       "0  Brewster's Millions (1985)           Comedy  \n",
       "1           Beach, The (2000)  Adventure|Drama  \n",
       "2               Patton (1970)        Drama|War  \n",
       "3        Wayne's World (1992)           Comedy  \n",
       "4     Bonnie and Clyde (1967)      Crime|Drama  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Add the latent factors to the movie information\n",
    "inner_raw_item = pd.DataFrame.from_dict(train_df._raw2inner_id_items, orient='index')\n",
    "inner_raw_item.column = inner_raw_item.rename(columns={0: \"index\"}, inplace=True)\n",
    "inner_raw_item = inner_raw_item.reset_index()\n",
    "inner_raw_item = pd.merge(latent_movie, inner_raw_item,how=\"left\", on=\"index\")\n",
    "inner_raw_item = inner_raw_item.rename(columns={\"level_0\": \"movieId\"})\n",
    "inner_raw_item = pd.merge(inner_raw_item,movies,how=\"left\", on=\"movieId\")\n",
    "inner_raw_item.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeUAAAEECAYAAAAWKCxFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9ebAmaV3n+8n9zXzzXc++VldV12Z30wWtbAqOLehAiGjABQkUMWCu4b06d4yWsWMucg0CBe51NMYFZ4Qbjk5cREFxcJhwFGHUQGyabnqleqvl1DmnTp3l3XPfnvtHnvPWOdW1b11VnZ9/qt4875v5ZL755vf5/Z7fIgkhBAUFBQUFBQUvOfJLPYCCgoKCgoKCnEKUCwoKCgoKbhIKUS4oKCgoKLhJKES5oKCgoKDgJqEQ5YKCgoKCgpsE9aUewNXw2GOPYRjGFX8+DMOr+vytQHGOtw8vh/N8OZwjvPTnGYYhhw8ffsmOX3B+bmlRNgyDQ4cOXfHnjxw5clWfvxUozvH24eVwni+Hc4SX/jyPHDnykh274MIU7uuCgoKCgoKbhEKUCwoKCgoKbhIKUS4oKCgoKLhJKES5oKCgoKDgJqEQ5YKCgoKCgpuEQpQLCm5yhBC03Yh2KNF2I4oeMgUFty+3dEpUQcHtjhCCZ08PWOr4LK45+HqH2YbJgckKkiS91MMrKCi4xhSWckHBTUzHi1nq+Du2LXV8Ol78Eo2ooKDgelKIckHBTYwbJpe1vaCg4NamEOWCgpuYsnHuFabzbS8oKLi1KUS5oOAmpmFpzDbMHdtmGyYNS3uJRlRQUHA9KabbBQU3MZIkcWCywni1hBl1OLSrQcPSiiCvgoLblEKUCwpuciRJolnWaRqCZll/qYdTUFBwHSnc1wUFBQUFBTcJhSgXFBQUFBTcJBSiXFBQUFBQcJNQiHJBQUFBQcFNQiHKBQUFBQUFNwmFKBcUFBQUFNwkFClRBQUFBZeIEIKOF+OGCWVDLXLGC645hSgXFBQUXALbO3ZtUXTsKrjWFO7rgoKCF7HVw3mx7RU9nDcpOnYV3AgKS7mgoGAHhUV4bi7UsauotFZwrSgs5YKC25zLtXoLi/DcFB27Cm4Exd1UUHAbcyVWb2ERnputjl1nX8uiY1fBtaQQ5YKC25jzWb3j1dJ5BbawCM/N9o5dRfR1wfXi5f0rKyi4DbhQms6lWL1nf75uqoVFeB6GHbtexh6DgutLIcoFBdeQG5nHKoSg40Y8c3rAIEhQZRBIO9zTF7N6z+fe3j9hFxZhQcFLQCHKBQVXyLkszOdWnRsStbwlpkdODzi65iAB802LZlnb4Z6+2DroxdzbhUVYUHBjueHR148//jg/9VM/9aLtX/3qV3nHO97Bu9/9bv7sz/7sRg+roOCy2BLFRxc6w38fPdlluePteN/1ilreEtMwTvPxACfbHkmW/32723qsYjDbMJmul3jVfH3HJOFC7u2CgoIbzw21lD/96U/zpS99CdM0d2yP45iPf/zjfOELX8A0Td7znvdw//33Mzo6eiOHV1BwyZzLwjy65lC3dJSzprrXMmq57Ua4YcIgiBEIDE0Z/k0AYZJi6SplQz2na1qWJBrbxnIlQV1FqcmCguvHDRXl+fl5fud3fod/+2//7Y7tR48eZX5+nlqtBsB9993Hww8/zFve8pYL7i8MQ44cOXLF4wmC4Ko+fytQnOP1oR1KLK45O7bFmYzTk5HFGStTAkZEFWc1veg+g0zBizMsTaYkv/j9JzoRDy0+gQBUTWNtENMsASG0nBAk6OkBVkVj9WSbhUzh8eU+27OSl5bAb1d37t8TLLU9xOZ4Z5sWqyfbrJ5nnMvui98/U742ovxyuF/h5XOeBZfPDRXlH/7hH2ZpaelF2x3HoVKpDF+Xy2Ucx3nR+87GMAwOHTp0xeM5cuTIVX3+VqA4x+tD243w9c6ObQLBmK2z4ZxxV1/KmvKWRdvq+KCDD8zWd36u7eaCPDs7B4CEoGTHqKrMrKHghCnjts7BqSrNso4kSSy2PWalwYuONzpZYa5pDV8fvAzLt+1GLC90mLVGdmyfmG9cE2/Ay+F+hZf+PIsJwc3LTRHoZds2rusOX7uuu0OkCwpuNs4VQDXXsNg/YdP1k8ty7V5KLrEbJjssXoFEs6wx07CwDAUh8gCR7cc62wUtIUgyGAQxbTcaju1y0nyKwiIFBdeXm0KU9+7dy8LCAt1uF8uy+Na3vsUHPvCBl3pYBQXn5UKFJC5F4Lavy7phgoRAsFO8twtd2VA5W9oFEuMVg7VBeM6I7+0TBwlB243xogRZhlPd4Ioiw4vCIgUF15eX9Jf0V3/1V3iex7vf/W4efPBBPvCBDyCE4B3veAcTExMv5dAKCi7KlRaSODsAaxAmpElK1dQIkgxDVVDlnULXsDRmt7mcIRdfARe0srcmDmv9gH44YLpsIm3K+8Uqe52LotRkQcH15YaL8uzs7DDl6W1ve9tw+/3338/9999/o4dTUHDDOdtdbesyT64N0FSZfpCgKzKvmq9TN8/8PCVJYqYsMTHf2GGZny3IW2xZ2VsTBzdMqJzDmr1ct7MkSeyfsDF1hZYTMmIbzDXMK46+PjuSu6Dg5U7xKygouAjXOgXo7HVZJ0xZd0Jqlk6YZGQCFtseXS+maRs73nu2ZX6p7uSzXwsETpjihDELLRcZKJe0HUVFznW+QogdBVLabowfpVdUIOVcKVt4goNCXNK+itSsgtuRQpQLbluuxUP7evQW3i6QEgIniMmApqWRZYIgyVjsBpzuBzRtY3ge7VDaEaAFl+5O3v4+geBUx8fUFR4+3mZ9EA2rgc3UTYQQnOwEhEmKoSrsapocmKoiSdIFg9IalnZZ1/uc+2p7dLz4ktbki57PBbcjhSgX3JZcq4f2lXRZuhhbArnc8Wi7Mc+sDHD8mG8cbTFaMUjSDEmSWGh5TGwGci13AxbXHDy9TcPSmagYQ8v2UjoXbQ9MO93zyTLw44ST7fzcTrY9qmaNk22P0/2Q9UE4zEPu+zFj1RIjtnHe6GsnjFnrB8NrJSEYsXOrXiAhSWAb2kWbZQguzaV+Pb6XgoKbgUKUC25LrtVD+1qkAJ3LYj8wWcHUFf75aIu7Zqt859SANWeA1/bYO1YmzQTfOdUjSFL6Xsx0I6+Ct9zxeWKxx90zNRT5TPOJcwWcneu4W+vLihwQxNmZ95JXAxNkHFnpoakKmiJjqhLPrg7y40xJlHWFcyEEOwS57cY8vdxjpJIHmY1WDKZqBrON8gWbZUhcWiR3kZpVcLtSiHLBbcm1emhfbQrQhS120DWFIBaM2hp+ZJIJMDUFSeTWpBMkrDsRVUsnzmT6TgScKad5vonGhY67NfbtJTpVCTRF4lQ3wo9TnDBFiC1bWXC6H+BGKbP1EjP1EsvdYMd+t1cWTTJY6XrIssQ/PLeOpSscb7l8964mEtKFm2U0rUuK5C5SswpuV254Q4qCWwchBG03YrHt0XajzYf0rcG1emhvCcd2LicF6HwWe9uNaDshx9YcljsBgyClXFKpGirVko6iQCrANhQkIIxTwjS//hJgqGcE9VwTkPMdt+PFw3OyDYVRW0eVoG7pPLrQoePGuEFCluXHWh0E1Ex9aCEvdQPGKgb7JmyaZY19Ezb7J2zKpTPXI0oSJFniG8daZELgRgllXeHkhkOSnRnvlkv9Vbsaw39nytIlLS9c7fdSUHCzUkwrC87JrR5Ic63yabevxTphPKyctSVuF7sW57PY1/sBa/2QiarBSs/HjxJO9XwOzzX45okWI2WdqZpJx43Y1TRRVRlDkYjIWzSqMnCOCl1sju3EhkOaiWGP5e3j2Z6/fGAyxgkSnlzsMtOw6LohByYrHG95jNkGuipz51gZRZY215jz+8LfdH1vRV/vn7CH11tVFE73fJq2znMrA1IBCy2XN+4fJ0wSMiFYbHs7XOpblv756m1f6Hspoq8LbicKUS44J7diIM3Za6j7J+xr8tCWJImGpe0IZIJLm6Rst8y30pCSOCVKU46c6iNLEook0bQNGmUdRYK7pqpsuDFLXY+VXsD+CZuDU1Uakza15gitzdraZ1fomqmXkIDFrs/6IOS50wP2jNlM1w22nGJb3aO2rpNt5EI+UikxCBM2nAhFljg0VWXUNmg5IQ1bHwp7ksGaE+3Ied66L7ZE8sSGw/6JCv/9qRUMTcWPElRZZrHlct98gxdWB8P9Xc1E70qLt1wKRbpVwUtFIcoF5+RWC6S5kGV/LcZ7pZOULYt9seOx0vUZBAnNssHTSz06XkSzrLPU9VnpBsw2LUqazHStRMuLGLdLZMD6ICRMeozrAd89rnLvXJmljsdi26NmnTn2M6sOliaz0gtoOSESEg+faPPq3U0mayXmGhZ1Ux1ep61a2IYKaSaoGDKjts6GE5HFGU1LY8w2yLZV3a6UVGRJMAgTwjjF0BRsQxneF/m1sFnquOwdq3K655GhIQN3TlSI0wRVOTPmm3Gid6t7iQpubQpRLjgnt1ogzfW27K90krLlZjU1mdVegBelRInHYsenbKj0gwQ3zMiAiarBcsfnZJanLJV0hcW2z0SttBkdLVjq+oRJymLbY82JWO4FjFV0do+U83XnOGXDiQAJu6RgGhaGKrN/osJ80xpep60I6ZNtDxDYhkqzbDDbKFGz9GHHqVpJ4bk1l9Wez0TNZMzW+O9PrrK+GXAGMGbr3DffGL5uWBrzzTIvrLrUrRpxmjFW0akZKrr64uUDN4iH1zLIFMRFiodcbyv2VvQSFdw+3JxP2IKXnFutxvH1tuyvZJKyXTw6XsTGIESSJDRVJtsc254xG0OT8SMD21CwdBVZAt1QSbOMSknF1OTN4C4JJ0wJIo8oztdlMwEr3YAsg4lqiSTZnuYkoUhQM3XkzW5QW9cpyfLc5K3QsZqp5W0gG+VhIRAhBF97Zp3HFrub68k9Dk7ZVHSVDaJhHrOlqzs6WEmSxCtm6yQZHF1zMDQZU1PQZQjTDD/K0FV1c10cVgch3dW8VevSch+zOTivVbplxS52PJwgoedFzDXL3DtXY8Q2rok4n30vbXkUTmw4gF24sguuK4UoF5yTWy2Q5npb9pc7ScmyjEdPdjm65qBrCj03JBMCS1eIkowDEzYLGw4SEEQZ+ydsaiWNRtlAFVCxNE62PVIRIpEHd3ndAVGcUjI1jq71mW9aLLQ9hIDTvYDvviO3Vpe7PqoqEycZ8w0LTRbD67D1b56TfAZDV6kYKmVDHU5iTra9oSBDnsv88PEOr93T5O6Z2rDilyqDF6Vs77AsyzL37Wqwe7TMMyt91p2QXpLy+MkeuqpQKSnMNcvsm7BYd6Jhk4ytBhvns0o7Xsxix2Op7fH8qkMvSHjkZJeuG3HXTHVYeexqOLvi2pZHYc+4TduNb4gru1jTfvlSiHLBebmegTTXmguJphCCjhtxuh8gSxJjFWN4Tpf64LucSYoQgieWuvz9s+tDQauWFPpBvOl+FhgK7BmzCeOEqqnx7ZMdDkxV2TdmAQqCjDtGypR1DYGgUVZxO7B33Gal65EIUGWJg5MV4iTjjrEyE5US0/USG4OQR092URWZhbZLvdwYNrfYuk4LLW8zAxlGbR3byFOeyrpC241ww4SljocY2sM5ipI3zWiWDSw936fg3JOfrV7NfpxhaipH11wato4fp9imihvGZOLc1/t8Hg43THDClKVuQC/ILdpMwMmOR8VUMQ0VebMwyZUK2fZ7acujMLLtGl1vV3axpv3yphDlgtuCs0XT0vP83sWOR2sQ8tRyj9VBNLQ675muICTpRUUwLvTgu9RJSseLOdn2d1iiUZrhRSmVkkZ1c9sjCx3uu6NBGGdM1S3aTkTjjiYtN0JCRpZhsmaQZDBdNxkRVe6dr/OEDM+tuqQC0jjbFFWVckmjF6Qoisyh6RphlOSBYk7AYsdnvmkNr9NYJXeXrznRUGxqlsozKz1WnRjbUEhSQceJaNj5+fpxRpKmTNdKhEm2I4J6OPk5a5LjhDGDMMEJYrw4xVQlvDBlfRBhqAoLLY++n1csk7aJ/4U8H1GcEqdn3PSSBLoqsT6I6HgtRjabeFypkG2/l05sOOwZtzfzxV+cWnY9KNa0X94Uolxw27Almg1LG1oagzDh6eUehiqjyhKpyC0fu6QSJNk5U3su5cF3tgDVTZWun+CGCYMgxlDloSUKECeCkqYwahtsOCF+lOCECa6fYJsqQZyhaQobg3CYEwyb68Iy2CWVEHhu1UGXJV69u8Fy18fQVGxDYa5hDVs5SkjYhkLPi9hwIiQEjy50ONXxmK6bSLKEbWi8bu8IXT/BCWPaTshaP+TJ5f7Qep5rGOybqLDU9XDDFCdMuG++QZgk1KzSjvrbwIutu3oJP045trmuvNoPMDUZP0pplPW8pKaei78TplQMFYkLLws0LI294zbPrzmb3znsalqEcUaXiOnGmZ7TVyNkW/cS5C7rs7meAY+3WuZDwbWlEOWC247tlkYYp0Rpxul+wMHJCmmcJ/gMgphM8KIewxd68A2FOIhZHYR0vHwtVCDYkmCBRJoJ1voBc02TxXbelUkIiYqh0vMjKiUVVZHQVZkgSXnupEPfz/ONa6ZKxdAo6fIwGCoD2k7AQ0se7tIKApio6Nw9U2OkYuxo9LAlFk6YDgXZDVJO9zy+s9LHCxLmRiz2jpWZa+Z1qAGeO+0QJPm1USRwwoR1R2GmbnBg0max41MpqURJxrqT0A999oyWh9eq7UYvsu4W2j5+nDDftFjueOxqWhxdd2iWDVQJ5poWiizl3akaFmVDZURUL+qteNV8HUWCb5/s4MUZSZIxWjHIhBha/Zfyfe74Ts+zJPFSBDzeapkPBdeW4lsuuO3YbmkYWt5YQQiINyOTJaBS0gi2RSpvcb4HX5ZlPLHU5WTbR1Nknl3pUy/rzDRM3CBhseMxVS1hGipZlmBpKqoqcfdMhZVegKUqhGnGPx1rYRsqkxWD798/xj8+v57XmUZivmZy5PSAhqlRtXTCOGO+abF7zGKtH7HQdhlp5qUlVwcRdttn16i9Q3S2ROTxxS4g8OMU5HxN/alTfSxNpePHyJu6Y+oKLSdkECZYmowqQZwJjp4e0CyH1Eoah+fqNMsaS50gLy4igabKPHyiw/fc0aCx2eTibMIkJYgzxm0Nu1TFDWIOTVXouRGjVXNHtbEti9ZZTS/qbpZlmcPzDXaNlIdxAqYm89yas8PFfKHvEy5t7falCHi81TIfCq4thSgX3DRcq4jT7Q9i21CYrZcIwiRPRdoSuqZ5zjXlcz34hBA8evJM4FaUJAgk+hsOEoLTvYCnTvWZrpc40fLYM2ozYWvcM9cABBuDEC/McKKI6VoJP0qpl3XGbY29YxXaXoQkQZpmnNgIqM5WGLcNwjRDVUAmY8ON0FSdkiYRJ4JUQBynrPWDF7nQy7rCPTNV/CjGixWeXu4hAae7IU1bIAcJS22fVOTrxIaqcGzNYayiM9sw+fvnNxACNEVmxNbpBzFIDAU5yTKOLA1oDULabsi9czXKuooXJcNobIGEoSpkQtByk2H6lSrB/skK8jZBPvu6X8p9IG1WQWturh8LIfDj7PyBfufY36Wu3d7ogMdbLfOh4NpSiHLBDeNCD9trGXG63dKQkJhtWtwzW8PSFGRZ3hF9PVEzL/rg63gxR1cHpEIQp7n1d3x9wL7xKsdbLostD1WRWGh7IOBEy8Uyqiy2PSq6SobEmhMiSeBGIXVL42Tbp2kbwzVmgUScgiIDSDy3NqCkKTznRqwPyhxd77PWj+gkDvvHbTQhCJKMI6cHVAz1RS70QRADMhIZQoJ+kGDpecUuAWiqxMmWg60ruTu8atDzI2RJYt+EjSJJTNdN7FL+iMjSXEI1VeKJEz3sks6pXj6hWel63DVdo+1GtJyI+abFSFlltKrT8eB/PrO+ecx8gqSrEvsmKueNkr6S++BFQrZ5Xosdj7YT0nKiF5X2vJnXbm+lzIeCa0shygWXzNVYshcT3WsVcbo1xrKusG/CRpLYsea6/T3bG0xcaF8nNhzW3JCeFxIlAlOT2Dtqc7Q1QAg4vu7yhjtHeepUD1PX2HBC2k5EnGbsHimTZflarRAgSxLTNZMR2+DQRIXxsk7bidhwQzRD5o6ROh0vwNZ1fPJKX0dOdRi1LZ73e5zqJ6x2PX7o7kmcIKJSUpBIGYQZJ9ZdDk5V0RUIk4y2G/Jd01VUWeahYxscmKry5HKXuqnT9yP2jVf49sk2aSbYNWLjRgmaIuMGCXvHbaoldShkB6ZskCTWBwG6qvLcap9MgBvmQW0zjTIzDZOapRNFCSVdY30QsdTx6Af5/VJSpc1pg4QsScw1rRdd8yBTaF3hfXC+QL9jaw7zTYtmWUMgDfd3tQVhCgu24HpQiHLBJXG1luzFRPdSrZZzPRQvNsa5hvUii3yx47Hc8YeW3Vaw0db5bN9XkqYstT0WWh6ZEHhRiqUp2CWFckmnWY5Z7vvMN8sstD3KhkbT1pGQ8KKEiZrBUifIhRmomhojts54zWSsWsKPYlYHMW0vRKTwdCdk16iGlCaUNIUwVnniVIc4yWhaBpau8Y2jLUbKBt9e7HLXdI2+F7HqxPhxyr7xCiVNJkMiSlPsksJ4zeDZ0wN2j9jYhoypq5zu+4xXSsSp4IuPLTFilXjFbI3ppsmpjk/N1FFlwYitIyOomyrrg1zsJSSqJZUoFaw7PmmWIZEHsw2ADSckFeBGGXEm6HoxVVNjw42pldPzCp8XZ3AO7b0c6/XsQD9BHnFfNWubnoh8f7MN87LWbq/2N7D93r2UcqIFL08KUS64JK7Wkr2Y6F6K1XK+h+LljHHrPVvRyXDmgb39vdv3FacpTphS0mTWN7sotb2YsYrBat9ntm7RckN2jxisOyFzzTKWJjPXLONvrrPeMWrRcSIOTFWZqurULY21nk8mBH4C3zzRHjaH2DNWJk5jxioWfpQMBU+VJXRV4XjLZddIma4fEyaCv35qhdfsHkFCoCkSJ1ou+yfLTFV1gijjm8fbVEwNJ0hJ04DSqMULaw79IEGWJAZhiqWpaJqEH6e4fsyeiQqjtkbHi/inFzbIgI4bo0i5SFuGQhhnSJLErpEyYZIOr3kUp9QtnY4XEScZuzYrj8VpRklVGLf18wqfpcn459h+9v1xIYv17EA/YLN2eDoseFI21Mteu72a38DZ9+7FyokWvHwpRLngkrja9beLie6lRJye76E4IpThWCQEaSYIkpQ0y/Ng3SAejnHrPMI4F5GtSOKeH1Ez9eF7t/aVZOBGgpIqo6sGQkiUdIXjGw6GouSpTqbKdMPk8GyVw3N1FBkMVUORBcthQs3SKRsK6rRCzVSwNIUX1lwW2j6GJnNsfUDZ0NAVCVWRiOMMy9B4bLFDkglKmkycpnT8lEoZapZOECVMjpZZ7OZVp4I4Yd+4jRumdP0IWWSYJY31fkCcZjh+xO5Rk/GKyWLXx9QV+kFCpaSz7riUdJm+n7A6CFAVmb1CsD4I+JunV1EVmeMbDk3LoG5pGIqEKqvsGSuRJClZBqosMwgTbENh77hNz4swNIV0s/LYockK03UTS1c5eIFSmCU5ZbZ+4fvgXJOzmXqJ8WoJL0rJhNhcWc/ztUdtnZaTFys5e3+Xs3Z7Nb+Bs+/di5UTLXj5UohyAXDxtbKrzZ28mOheitVyvoeiF+epTaYm03YiXthwNoO8YNwu0bB1ZjerWW2N19CUzUhiwbHTA7woRZF8Rjbfa+nKsOZxSZPo+wl2ScFQZXRF4q7pKpNVA9tU2TNW5u6ZOq+cq/H8mjvswtQPU1QlP46u5+uzq4OYkpqysFnxK04yglgwCEL2TdiokiAAjrWcPI85SAjCBENTuWtaR1E1wsRn91gZIQTT1RKKJJiuW6z2fBY6PmkmmK6P8PfPrtIPUixN5uBkFbuk4YQpp7oBFUNltm6CEGwMfPaMVTjdC5BlCSeMsXSFJxe7HN9w80lAnPHs6oCmlQuYrspYukw3yphtlKiWVJAEY7bOPTP5dcg6HuMVnUGQUDd1yrrCfNO8qAgdmKwwXjF2lEXdzosFTvDN420mayaKnGeOb4a4ISEx0zC5e6a2o9jJ1Ub1X8r27dzMQWUFNxeFKBdc0lrZ1eZOXkh0z54QzDbMcz40z/fwszQZIfKuSS+sOXx7qYsQUC1pBEme/rN3LB4GAW31Nx61df75eJuGpSMhGCnnLteOGzEIYlb7eZtEU1c5OGXTdkNmGiZBnAdwlRQZXZe5e6ZGpaRxqhcyXjEYtfX8erY91p2Q4y2HO0bKTFR04jgl3mwGoUoC01CxdDlPH8pSnBiWui5pCk6UIDLoxylVU2a0nK8nv/qOBo+caPH0aQdDlvm+faOcaDs0LZ1GWWfM1nlhLbe+u37KiF3i4YUOuiLTKGscmLSplzQGYYKuSnzXdA0vTNgzVkZVJO6erqNKGUdWB7yw4XDfXGO4v6qp0iirRCmMlg2iWHBsY4ATZUxUDTb0kKVugGUo7Bu3KWtwdN2n44VEaUKjrF3SWuraIOTUZrraUsffcT+eLXBOmLLuRDTK+mbHqlyY918gwvtKuJrfwIUEvQgeK9hOIcoFl7RWdi1yJ8/lKryUCcH2Slp1SxtW0tp6r+i2N7sH+ZR0mZqpk2UZ8mY1rK4X4YS5KA/Po2LwcCaYrJVIU0GQ5K0XEIJnTg/YcEKSNEORJZ5fcxmzdabrFneO2dRNjRfWXY61PRRJ8Pyqy2zTZKSsI5AYtTW8KMFLMp5ZcegFMU8u93jDnaPULZ1RW0eTBP0w45+Pr6HIEkGc8n13jhFEAXdO2PyPJ08Tp4KyqTBZNakaKm4YMlKzeOVcleMbLodn865Oz631GXgZqqwwCGIGfszaIOCOEZu5RomljoumyGiqTFlTN13xMpErKKlg6TLzjSpIMGobBFFCkJgcW3cZs0uEccb+ySonNhyqpoYTZZRUlSMrAxbaHntGy6z0PBqWyko3pOXk9aedIO/o9OhCh0GYokiw3AnIsoypRnlHVHyWZSx2fE46EmKlz+mux/a4+O3349kCF8bpZmvLM9W8LhThfaVczW/gbEHfKidaNyQjazEAACAASURBVNWi+UTBDgpRLrhk19r1yJ3cPiEQ5DWQH1/sYuoK85sP1O0PLYncRdq0z5SXfKabj9VQFXRVoaRJeKFE14tJM8FI2aDthMMobEmS8qLJSFiaitg0dFpuTElTyERe+EJTFV7YyNOeUiHhx4K2F5MBx1seUZqhKxItx6flhrxyto6qybhRjBumbAxCEpGBBBISS12fmqkxWTUY+Bb/9O0lMgGWKrNnzKbnR3zXdIO/fHSROydsul7Cc6sDVroBr987iqnLTFdLOEEesGRoMkfXPCxdQ1czFjsefpgxWTMI4ozVvs++iQodL0ZXZaZrJQxNpeWEVEsqdklhz5hNlKQ8fKKDpiosdXxefccIuiqYrJksdlycKOFU12PfeJV94zaqJHGs5TJSNhAiL3e5a6TMUjcvZPL6PXkTx0EQ87Vn17E3RTQTgkcWOkxUDU71Y6qGwoit07Q0njo14MhKn/VWi9KphDtGysw3S2TbhHnrfjxb4EqazB1NiyhJhqVJBdIlL61cDlf6Gzhb0LfKiRbNJwrOphDlgpes1q4QgtM9n41BQIag5UT4cYYiweOLCn6UMl4xzlo/lNhwYuZHdpaXLBv5w3iykj/0HlnogATTNZNdTZPFtkejbAw7JblhgirnHaO2qk1JQM3SUDd1oGFqZJkgyQR+mDDfKCFLsNj2aDkhWZYHFD2/5iA2q2NN1QymayXW+iErPZ8oyYiThJjcnV63VAxdpWqq3DVZJUhSBmFCzwtRZJmFtoNZ0oiSDCdKmKgaTNRyF6mU+FRNhYVOwDdPtKmVNI63XOYbJqMVgyjOsA0FRZI4OFlltedRMVTqpsZs08I2VJ5bdREiI87yIieLbY9BENEom/S9iLtn6/S8kIOTNrtHTdYGIUGS8rq9ozhhgiJJ9IKYasngVNdnvR+AJG9GX+cejZVugF1SidKMJM3IRN5eIxPgRglRJojjlHaS8dhijz3jFl89cka8FUXmkZMdJmpTaNtKWW/djzsELohZHQQ8vdRjvR2f6QI2U73sspTX2428XdC3yokWa80FZ3NDRTnLMn71V3+VZ599Fl3X+djHPsauXbuGf//Yxz7Go48+SrlcBuBTn/oUlUrlRg7xZYcQAoTA0hUGQTK0Mq5Xrd3thTvaTsh6P+C50w6LXZ8syyhpMs2yga5ILHe84QMdGEZDh0nKWs9HbOYMB5lC3VSZaVhIuBzSbCollXJJpawqLHZ8EpGL/r6JCgenqpsVn/JmCFWzRpikGKrCwUmb59dcJASGqlDaXK8eq+h4UcpqPyCIU071fKZrJVb7Ead7AVVTw4tinjoVECYJmqLw+FIXP065o1lmvmnScX1WBwYtp0WlpLLU8RFklDSFJIP1gc9UrcTJjQGv2j1Cx49pWCXiJKFq6nQ6Dm0vZbntsnfUph/mNayXewGHZ2uEScaxdYeyrtL1I/aM2UxWdColjRfWHdIMdAXmRyo8udQlTgQbfsRkLRe3mYbJIIzZM1ZG0yTmm2UGQYwsKSx1XUZtEz9KmKiWGPgJp/t5HWohMg5NVVkbhPT9iMeXe0iKRMPK16DTTBAmCRISE7X8u9U0mWNrLgLoeXlzkH6QYIm8jrVtqAyCmGb5TBvGc5XMtAyV7mrCdNOiWk4J4xRVlRmvli5LUF+KHsbtzdiFNBM76oBD0Xzi5cwN/ea/8pWvEEURf/qnf8pjjz3GJz7xCX7/939/+Penn36az3zmMzSbzRs5rJct2x9EAoEXp4zbOgenqsP11+t1vK1KS7ubFmVDJhUpLSfGMhTKJZXjGy6jdomZep6HLCGG0dAgWOn6NMsGzbLG4nIfs+mwb7yMEyYsdgI23BgnzIVAkaHtRGQC2m6LfpBQM1VGyhptN887tvQ8wGyuaeHHGQstj8WOx2zdpOfna6OLnYCJqs5M1WL3iMXJjk/Pj2iWdSaqRr6mrSoEkWCQBtw332BtEOYTCwkkSeEbRzco6yojZR1Dl3lu1QXyrkx7RiyOrw947d4xnl8d0PdiOm7EfbuaeW9kRcPWJUqanFvcqqA8V+dk2yNMMoI4RVMVlrsBmiLhBAm9IOH5NZcDkxWm6ibPrw44vj6g5cWs9/PSn5NVAzdKaDkRpi6TZRkgsWvEIhPw7cUek9UykpSRCDjdD4YBeZnIuGuyyomNAZIks6tZxosSnlzscueYyavvGOGRky30RKFuqeweKYOA9iBk3cn3UzMtZCm3pOMs91mM2vowgG57kZizhbOkyThhTBBnGFruDpeQcKOUCz1FzhZ3hLhhbmQhBMuuYHmhgyD3Flm6Oqw4VjSfeHlzQ0X5kUce4Q1veAMAhw8f5qmnnhr+LcsyFhYW+MhHPsLGxgbvfOc7eec733kjh/eyY/t61lY1Jn+zIMTlVii6mLtPCMHJlstji11kwI8TBIKFjgeSxP6JCsckl/GqwSBICOKMk22P79ndYLZhstDyhm7mqqnS9xOcMKVq1oY5n6au0PVipjf7+C51fNYGIQcm7LxRgiQYBBnPrPRJBZvtCy0mKgbWZi/f5W7AeMUgyzI6XkRJk3h+1UGSJMZsQcVQOdX32TtapmHpdPyIjUFIGGUYqky5JONHCSVNJhUJYxUDTZEpqXlLR0NVUBWZp071mWmUGLENDFVmfRDSC1JW+xGDMK94dW+9xEov5IW1ATVLo1mS+frRFsfWHVb6IaNlg/mmya6mhZAl6qZGzdQwVBmEQJUlMikXriTNWO36KHJezatuatiGykzN5FsnWnhxxnjVYKUX8vRKn/GKwWovRN8cux+nyICuCla6HgenaszU8knTC6s92l5KJhJEWaJp5R24hKTQGuR1sdMs76+cphkn2wFhmnF8w2WuYeGFMa+cq/HNEx10SUZCcHiuwcHJCrJ8Zk15e3tIgcAJEp5adskyyDgj5jMN87K7Q1m6Msxt3s71cCN3vJiltsesNYKExHTDxAlTZhoW49VSEX39MueGirLjONi2PXytKApJkqCqKp7n8ZM/+ZP8zM/8DGma8r73vY+7776bgwcPnnd/YRhy5MiRKx5PEARX9flbgQudYzuUWNxsFr8dM+rQNMQ5PrGTZVewtG09drZpMVM+98Nk2RV8c9HjmVUHJNgzWqHrhoxWNAZBQhhndAYxpCFBnDBeSjEUieWlJZqGQI8V9MTBUCTCgWBjsxnCouyTRTGLS4ukrs3prsMg1Wi5Ibausxx4pJFG5Dm0AoWlbpBb1H6EkTj0NgTKTJWWn+w4l4lGhag3wEGl03ZRNJ2B42EIHZFmbHRTukFeoGO1n4u/LstM2DVabsBExeBUy6HlhZQ0lbunazQthbWOQ6wZrHQ9ZAQLbY/pusmJDZe5ppVHjAtYGwQoUolGSUVTYL5ustJ1+ebxDpN1i7KusNxxCeKI7907xkprQC9I8IMIN8mLU/Q9MDWJ9X6IEII4ybtTjZZ12l7E4qqLKuXdoOZsgziOGS2rLK33WRkxkGIXRdaZLCs4oaDlBGx0feJY0O45vOAG3DNd5c5RC1P1cxe0H7HcCXML1pN4YrHFfLNM2wnoD0xquoJtaOhyyvfeUeFUx+PY6R6alHHfjEkWq8yPKExLbZ59tr3jHuqEEiudkDARBKnE2iDgZMfnzokap7setga9rsSYPsaR57tYmkxJzpc3vDjb8frx5T6SBLKqESYCy9BwvRBVPtPOUwJGRBVnNeVa0g4los17djstaYDoClav6dEKbjVuqCjbto3rusPXWZahqvkQTNPkfe97H6aZuytf+9rX8swzz1xQlA3D4NChQ1c8niNHjlzV528FLnSObTfC1zsv2n5oV+Oi1kHbjVhe6DBr5ZG2AsEgTJHqlRfN9ttuxDPPraOaMtWKjABaoWB+vEmSZrxqxub5NYd6LUNTJCZrJiN2nvpyaFczdw+7EamRj3UQJoSqgwTMTdU4dWqJudk59o2XWX1ug2dOtBFCRQ5T9k6NYhgKNc0gSTPurtVRZQnTEozUTWRJIrNtJClizmoO16xlVeae8VEW2h4v9CRaXsx4s0oYZ1QslQNTVZa6PmGUsnuswlQtd8EemLDp+DFfP9pismpSKhnM1k0GfszBqTJ1u8zjy13GayZVS2NaWKz3fe6csClrKiO2ihtlLLY9nDDFj1ImqyWiSkaQgpBUMiFRLemw6er0opRVN2G+UeYUCk2g44XsHitjaQJFsVhuueydrLLc9RkECQM/ZrphUSmp9AOVEVvHNvRho4klJ2O0UmeqUkK18taLvVjGNgSThkLN1BhrVNk9brPYcZFVjZXN8qWNsk7VNDjaDhirVymbGjEaZdMgliQWOz6qLFPSUmJJY60bMVk12VOp4HfX0KwqE/ONF6XOPbLQYW11HS/O1/Zn6hXqqUazalOv2IzbBhLgCUhRcISMJiSEBEKX8IHZusmIoTAnVYfLIQJwAtg9UcPSFZwoI4pT9o7b3Dtf32GtXwvabsTza08wOzu3Y/ul/O6uFbe7MXIrc0NF+VWvehVf+9rXeOtb38pjjz3G/v37h387ceIE/+bf/Bv+8i//kizLePTRR/nxH//xGzm8lx1XUwxhe9SoQLDc8Yf9ds9V7CGM8zzVMTuvD50JCU2GV82PsH+8zGv3NDcFIz1nsNn2sdqGwpidF4pQ5TM5n3ZJpe9HIECV8+5QigT7JiwEEqkQvLAZfQwyp7p5a0eFvHuTQAwrbUkI7pywMVWZmpUX2jAUmclqiamayXdNVxDAWj+gE8SsORElVWHViZCAg5M1JmolJqoGI5bGC2sObS9huetRUmUOz9WoGSptL+KRJEWWJJwg5s4xCy8KUGQJIXK3bCYyTF2hWdZ5Yc2h60c0yCO0/TSlVtZ4VbnBSt9j/7hN1dTyXOMoRtM0Vk/32T9V5YnFLk27xGo/YLJmsdzx0OU8it00VJa7HlGSced4hShOeGo5pLpHZdTOy4TWrbwUaGsQISkSuixjGiobg5gx20BTJBZaeY3rqqlyuifQFDYnEDFIgnUnwglSZhoaL6zlE/TxSgnTUFls+4zr2vD+2i5QHS+m40V5f+cNFyHy++zQZIUwzkg3u32daHsESR48F8QpJVXm3tkqGXmE+ELL48CkTZIxFGSAZPM/ZUPDj0Pqlk7Pi3hu1bnmwV4NK4+G306xjlywxQ0V5Te/+c18/etf5yd+4icQQvDrv/7r/OEf/iHz8/P84A/+IG9/+9t517vehaZpvP3tb2ffvn03cngvO66kGML2deStWsdbzR22F3A4u9iDoeXRznZJwTQskjTjwFSNw/N1RjYb1c+PlM+7Rn32WO+bbyAAL0ppiir7J2yeOT3ANlSm6waSlLfo6/oRvSDilfMNSppC3cqjqIM4QZJk5psWmiqx3PHxonR4PC/OeO60w4FJm5KqcGiqSpxk7B6zsDSVME4ZLWvESYYqSwz8mBNHHmf1b/6Jt/6rX8aN8iCxkq7Q8mJUWUJTVSbtEv0o2VynzoOsXn1HAzdIUBSZIMn45pf+mNrELNqe13DXVJWljsfxDY8oSXj93lFOrLuEWX7cjSPf4nOf+QKH/8WP8N1v/jGaZYMDExa/9+k/wrzr+3nkf36ZsLNC+i/fi6bInO4H+HHe6amkKaQIpmydMdvIS1LWTTQF/vm/fZb17/wzz42Pk6QZvjNg76vewOj3vJXFb/89jXqdt73lh9jVLNGeqnCq5yMEtAYRlqHgRwnfNVOj54V0/TzVbfdoGZFJebGWWJAJsA2Fsi5javJm0wjBH/7WR0ndLmunV9A0jfHxcabmd1OZ2sPq0gKv//H3o2+46IrEiGVwqp8vZWTkVmjTzsU8SjMWV1b5m9//FaIko7V8nNrYNM1qmdfd/y/R7nz98L4etXVKmsJi2xvejxeqT/3TP/3TZFnGsWPHaDab1Ot1Xv/61/NzP/dzl/S7mylLTMw3rmn61Sc+8Qmefvpp1tfXCYKAubk5Go0G733ve/nc5z7Hb/3Wb132Pk+cOMGf//mf88ADD/Cf//N/5stf/jIA3//938/P//zPEwQBH/rQh2i1WpTLZT75yU/SbDb56le/yu/93u+hqirveMc7eNe73sVgMOBDH/oQjuMQxzEPPvggr3zlK/nt3/5t3vrWt3LnnXde1fnfLtxQUZZlmY9+9KM7tu3du3f4/w9+8IN88IMfvJFDui25nBZxkiQNZ+hb1u/5HhDbA2QkBEmSccqLMFR5mB+aW7kM97dV7OHARJmNgc/aIEJXZPaP2+wesc5bnOR8QWRnF24YAZzVlCeWupxse2w4Ibqi8OSpHrahEiVg2RrLHZ+7ZursGlFxowQvymhYOupm1a+SJtP14+Gkom6qRElGmuVrvMFm/nTLiVnwXRRV5shKn5MtF0mSmKyVuHumxlPHwJAE98zWWVh3CbKMrhdx33wTPwyRZJnvnOoRpbDuBMRpldVel++artHre9QsHdtQ0ZW8GlXHi2naBjVTQZY1NEXizXdN8MRSl1TAI197hIM//JPY+++jUspbQhqqyrf++s/YO/Uaul6EmggeO9lhV7OMqUnMNy2OrTs0ygZxmuGFGXpDIdjsaNVyI0qayqt/6Mf4nvvfxnNrDq4f8g+/+wA/cPcb+f4f+hHmmxamKnF0w+PYuosXxgzCmHvmamwMAk71Qp5/foPX3zlKvaRRNSWiNMM0FO4wLGqmhiyDJktM103CJNuc1En8ykc/zoHJCr/7u7/L6Ogo73nPe2i7Eb/3h58FYKxiECUZbSekYWuUjNyLUNYVTnY82AzW0hQZxazy9v/j1/HjjP/22/+O73v3/8Z99xzklXM1vr3YI4xTDE3BNvK8eH17YvRZ9/F2/uiP/giABx98kLe+9a288Y1vvNBP8pxc60I8Dz74IAB/8Rd/wbFjx/ilX/olAB566KEr3ucnP/lJfu3Xfo3FxUW+9KUv8fnPfx5ZlnnPe97Dm970Jr7xjW+wf/9+fuEXfoEvf/nLfOpTn+KXf/mX+fjHP84XvvAFTNPkPe95D/fffz+f/exnee1rX8v73/9+jh07xgMPPMAXv/hF3v/+9/PAAw/w6U9/+ppch1udIhnuNuNyW8RdTn7mzupbWzm+GqMVnUbZuGCupUye91op6SjAHWNl9l/lmIQQtN2IpzsyxxfW0GQAiTjJsHUFU1OYrusgBBn5WrGiqxiqih/lqVDDILWGhaXn7fwMTQEhOLbubjZRyAuMaKrMSi9gpmHwwprDcsfLOzQlCf3TA2aSEEtX2TVWYbHVx9AVWr0Qf/E7/NEffx4UDdkoU3vjT+KvLBD+w2d5InJpvO6dnOpUWPz6l4hUE13KsMdmaLkRS12Xe2fqqEnIP/7Jf8D3PCxVYux1P05/4BAuPs0TS0+x9zU/xJsP/RxjZZ3PfPo/EXoOC1/5Y6yROYKTJ4j+7lMcS31G5vaxcexJ7F33kA5OsSyg05zgcXOcQ294G4/+w/+gHLXpR4I7ZhUeW+wAEp7TJ0tTTjkJ3t98jpPNEaLSGA//zeeRZJX+8gukaYpUn0KXUsKNZaRSha/fcYjg9DF+8F0f4KFHvo7XWcP3fb77TT/G7OE38sjffhF3app7Xv29HJqqUY4657z3GlZeve3rzzzNb/7KL+L1e7zmTT/CWOUtfO7f/zuWjj1LGkdIikZzZjfNmTsYm9tLa2WRu9/1AfLFCdAVCYTgP/32v+fOQ69g172v53c/+kscuvd7ePd7f5L/8MmPIYTgyLe/ie97yJKEJMH7fuqn+MVf/MXz/u7+9m//lrGxMX7zN38zvw6SxIc//GEeeughkiThAx/4AB/5yEfQdZ0Pf/jDfP7zn+fw4cNMzu/hNz75cWQJRpsNPv7xj/Od73yH3/iN30DTNN71rndx/Pjx4X5KpRK9Xo8Pf/jDvOY1rwGg2+3yj//4j7ztbW/jwQcfpFaroWk7XeELCwt88IMfpN1u8wM/8AO86U1v4tlnn+VjH/sYAM8//zx/8Ad/wCte8Qpc1+Vnf/Znufvuu+n3+3z2s5/lZ3/2Z/nMZz6DouSTliRJMAyDRx55ZGhIvfGNb+RTn/oUR48eZX5+nlqtBsB9993Hww8/TKfT4aGHHuJzn/scg8GAIAj41//6X/Pe976XZ5999qIxROfiWlrynufxwAMP0O/30TSNT37yk0xMTNxwS74Q5duMy20Rdyll/ras1hMbzo5CB2IzB3isYlDW1fOuTXe8mKVugCLLjG66BrteQtc/d7rJpY7p2dMDFloeXz+6QSybjNk6lZJCvawSxAaKLKErEgutgGpJxZjJA3ZUGfaO23S9eLj/XU2L+abF8mYTBIHg8FwdEMMCIwBelDAIY5IUNEVlpediaPkEYHk9IPEinljuoSoyy12PYxsuJ7/4B0z+6APce2APR/7+S3iP/zXGzN0oUoYyfYC5e7+Xp//fX2bqHf+OTLPwv/YfEUIwWjYYsXXaTsg3v/In1KYPceB1b2FcDfjaf/oI7/w/P8U/bDxNHEY0RiYI44y/+84q4695G/J//QLJ4XfAiYeQZIXmm/93RhWXhc99lP0/8r9y/O/+hFf/q19FmFWe+h+fo/et/8ZTe9+IF6Y4g4CZZoVv/PVfIIyvkLod9EqDmTe9HydVMeOM1V5AWU+J+y3+xc//3/z9F/8Lgyf/jpkf/SU2/uY/on/vDyIe+6+Ux2YRbptlfYYTzz7F//Kh/4c7x6s8+8TDvGK2xit+5v0ESYaxuUSwtrhxfq8OoGsqP/oLH2X19Ap/9bu/yuTh+2mtr/J//d7/xz/+9V+yfOxZ9h5+Hfe+8S08+fW/pRKXma7lDUgyIYjSjGPrLve8+vt48utf4S0/9IOI0OP0C09wz0yN0wsvcOCVryWOIz7yO/+Fu/9/9t47Tqrybv9/n3PmTJ/ZmZ3tyy6wLG0VFBBEAQti11iDQoTHFruP2AkGRUV87DEaS6wRJSKCRozGIKgYVEBRmtQFdtleprfTf38MTFjFkkSfx9fv6/UPL845M/e5Z899PvenXVdtNSveWsDjjz/O1KlTCYVC+723F154AVmWmTp1KhMmTGDTpk3MmDGDRx99lBkzZnDRRRexc+dOstnc87VmzRpOmXwxv7n2Ss67ajrlVX3Y9NHfeeqppxgzZgyKorBgwQIAxo8fzwsvvEBJSQnjxo1j6dKlPbpYtmzZwrJlyzj11FP3e2+Q61R57LHHMAyDo446igkTJjBz5kzmzJlDbW0tJ5xwAq+++io1NTX8+te/5pRTTsFms+FwOLjqqqsAKCwsxLIs7r33Xurq6ujbty/JZDJP8OTxeEgkEj2O7T2eTCa57bbbgFyU4cknn+Txxx9n1KhRrFy5Er/fz6pVq/5lo/xDevJvvvkmBxxwAFdddRWLFi3iqaee4re//e3/uif/s1H+CeCHpPf7V2n7vuv6vcavOZImrujs6kxRHXQT9NjyvMReh0zVnh7L7xKd/0/vKeiWiaQ12mIZdofTRNMauiWg6AaN4TSFnlyfrkOWUHULr2BR4JJxyiKRlIJFTo+42CPTt8hNWjV7kFOUFrjycwi4bEQzev7/lmWyfFs3AqCbJpqh59SoHDba4hmKZAFNltjantgTApao9pg02BykJT+qplBTN5RP3ngRW2gQmt2PUxJxGCncXh+De5ehGyZbvAE6Vi4mtfVjDJuLwb+4BLp3ku7ayqZdn7N70FhUzeDV/5mGpaRweHw4HSNZsKYJl03CLolIArhWPkukoxlLV+in7UaJJlGVDPXLXkZPRvjghQdx2CWUSAeGmiX+0cvYS/qgduykuSmOZer4Bo8n4yhF+/g5MisWYnc6iZsqtn6jCTkUTH85uiViCBKIErppYsTbKCzuRRST4r6Dafl0CbHNK6kZOZ5F91yDZZrYZJmW5mZOmDiVTLiVx39/Dz63g8ICH8lkkt8/+SxvvfN3suk0816ez+m/nEQ4raIZBmtXfURx7VAy2Sx6JkEyGuape28j0tWGJIrU1dVxeG0Ird7HO59tYe59v6G1vRNNVRAQMNq3MffV52htrKetrY2TTjiOJUuWsGbNGkaPHAGSgMvpZPzwAQTdMp0DBqCqKg6Hg0QiwS233EIkkusE8Hq9rFu3jk2bNqGqKtOmTWPixIm88sortLW1MW3aNDRN48abf8POXQ1kM2neeustGhobufeuO2jYtpnf33YtpRVVGIbOoP41AIRCISZNmoQkSYRCIWbPns3WrVtJJBJceumlPPPMMzidTgCeeOIJNm/ezPz58wFYtWoVkUiEjz76iFmzZgFgt9uZMmUKgiCgabnNaH19PbfffjsAnZ2dtLW1ccEFFzB58mTOOOMMnnjiCZLJJNdeey0PPfQQxx57LKZpkslkOOiggzAMA5fLxe23346qqhQVFdHV1fW1LptUKpU30lu2bOHpp59m5MiRjBo1Kn9NMplk7ty5vP766xx99NFcffXVPTz5QCDAnDlzehj7HTt2YFkWhYWF+Hy+/9iTP//88zGMXPtbS0sLfr8fAL/fj9Pp/Lc8+X8HP2yt/8/4l7HX6K1piPT419qHXvJfwb/KY/1dxyNpjeZImnBKY0dHko6EwvvbOulMaghYeY94b763qtD9NTawH+qe3HYp/xutbYrywZZOOhJZklkVy4TupIIoCmxqT9IazWJZJh6HneqgkzKfnXhGZ1dnmi2tcV5d08IXjTEqA//0vr+6MYKcqEJnIks8o9IRzaDpBtta40gIlPicCIKFqhuUFzgp9DjoiGfZ0ZXCLdtIZDTqoyDoCgcUWhR7nezctB7LU0SZ34FskxAAtz+AoaQJR8Js70zRuO4jzH5jKT/tBoK9B7P1vUXYfSGMbArvCdOIe6sx0xH6nHE9vj4HYIk2WmNZwikFv1tmRX0XqpLFUz2Yg487m94HHkr9W8/i6T0U0Wbn4JOnYveHOO3K3zLuwtuonnAeyE6qZgXZyAAAIABJREFUJkxFEMDttOPoN5LCA48iuX4ZnrK+GEaOEOXUq+/EsjnItO2gO5ljBGuPZ/DYRSxdJbPk92iJMB1LnqD86KnIqY5ckZuZxYh34PL6+NXtT5FVVD58cz4fbu3kpT/+gePPmsx9j/6R3v3rSCo6Kzbtprm5haoREzjrmtk0RbNEUioVfQbSe8ih2KTcq8vh8uINhjj5st/Sf9R4aoccQnHNgaQVA7ddQpZlZt//CBfdPJtMrJv+JR4WP/sQl0y/i6OOPYFdu3aRzWYZMWIE9913H8cddxxBv4doJMxlF5zHhAkTmDVrFmPHjsXr9fLEE08wevRo5s6dy5133skXX3zB0KFDGTx4MIcccgibN28Gci1HRUVFAPQ/cBj/WLGCkqq+HD/xAn7/h8fRdZ3zrrqZqn4DEASBC66/jdOnXs4ho8cCsGHDBm699VaeffZZQqEQsiyzdOlSBEFg9uzZeYMMcNlllzF69GjOOeccACorKznllFM477zzWLRoEc3NzbS2tjJv3jxeeuklVFWlubmZvn37cs899zB37lyqqqrYuHEjNpuN9vZcp3QoFCKdTgN7yH8aGznppJP46KOPiEQirF+/HkEQUFWVl19+meHDh2MYBv369aOhoYFoNIqqqnz66acMGzaM7du3c80113DuuedSVdWzHUxVVc477zxeeuklXnzxRQBmzpzJbbfdxty5czniiCN4+umne3xm9erVDBw4EABZlvOe/D333PNvefKQ486YOnUqL774Iscee2z+moEDB7Jq1ar9vpd+aPzsKf8f44dWifkmibhvarf4rraolKLv0z4iEPLacTtsxDIqo/uF8gIP+8N3SS5+n3vay9zkd8o0dCVpCGfwOm1YVi6U3JnIMriikLBi4rRDR1yhX7GHjGrQntCIK3EGlPpwiAJrm2JArijLJUt8sTtKbamX6kI3W9oS7I6kSWZ14hmNqqCDjJaTcbQQwDIpcNkpdMn0KfESSWapDNjpX+LBROCDLe10xhW0ls1E37iHiGXhcdioOP5iRpxxKR+9eD8+p500dhyHnUd7vB27lMt0iqLEqNMv5pP5D5A1JQRBxOnxEk6peMtqMXeuZ9Chx7JhcT2dix9AMhQkb4iw6UTVTeRgOW67jQGVQba2J/A6ZaKijZYta+nqM4xKpxvJ4SIayZFxSKJE4eHn8Nqjd2AYBqGAH8mWa3lyhzzoYi1tKQO3x0OHqlApqSR0jVS4g4VzrkJVFNzFNsBCFAQCbhnN76JbtiMedRXlXV/S9ven6Fq9GKUgiCiJyE437TsjJKIRXrr/twwc9ws2fbiYNe++Trh+M71qb2bd7gieXgOJpt9nd9ykpP9BbP30fXZvXsORE04krRh7FLcsZBFsooAgSfQ/bgp/uu8WtHg3Xp+HX0y+kK3tCcJJhdLqfmzuSNGpOTBMk46uCLFIN8/cfxuGkqKjowPDMBg7diyvv/46I0eO5NNPP2XSpEmsX78ej8dDQ0MDt8y+l3BKZevWrXzyySe8/fbbAHmvE3KFqo888gjbt29n1qxZ3HXXXdw663aq64YTWTCPo04+i7pho1j03KNU9e6L0+Vm0mXX87uZ1/DEXdOR7Q7mzLkLJRFBUZQ8v0D//v3505/+xNSpU3E4HJSXl7No0SIAzjzzzK+tn8rKSgCKiorIZrM0NTWRyWQ4//zzgRw/REtLC7NmzeLmm29G13V2797NNddcw5lnnslZZ53F8OHDGTVqFAsWLKCyspJ3330XgC+++IIpU6bQ0NDAhg0bCAQChMNhJk2alDeMsiwzffp0LrroIizL4qyzzqK0tJRZs2ahqiqLFy8mm81SX1+fp1m22WyMHTsWl8uV567Y15PXNI0+ffr0mGckEumRTlAUhRkzZuDxePKh8n299lQqhd/v/1ZPHnKpiPr6ei699NL8vIuLi/OblR8bPxvl/2P80Cox3yQR923V19/WFuVx2FB0g3/67QIuWaLY79pTBLP/7zVNc09FdCYngyhaFHsdPSQXv/OefA42t8WJpzTWN0WxgM6EQv9SLx5ZpMTrZGtnAqdk4HfntI6TWQNEAVU3UXQNr8OOaZi4XA4KnBJepx2BXLW1ppt0JxW8DhsN3SmawmnCaY3mSIrmqJNNrXGCHgeFLpmkavDRjk5G9i5idUMYr0Nie3uS4X2CaJqBx2FH6FtHevK9CIJAr6CbEp8dwzQRRJHDLroDp13iy5Y4pT4X3akCWhUN9+6VlBW4iNqGcODU29naliDx2m24qw4gpRhYu7cgF5SQMmwU9x2Ma9z5OCTY9NxvSMUj9D/p17T89Q+kshqCICCJAlVBF6l+o5E8hSiVhzCgxknjFx+SFD0E+x/Czu4k/j5DKa4dwvaOJEVBF5FnbyCR1UmnVPyiQN0xv8TvsGhe9TalJSEa3QWM/fUdVBYX0vTlamSXi4raIZhHHM4HWzsR0ir9j52EHHQRadOoGP0LIvWfI5QPxh7roimaofehJ9H6xTJ6nXApFia2j/7KIceewWfdO/nHyk/pVTeK9WvXoxkWa7c1IvhKGXHaVXgkkzfuv4KrHniJ9W0J1jbFUHSDqXc+QyKtMnTEoQw4aCRfLpmP5ArQpshkkkm6O1O0xRS8dpEirx23PwihPgRCxdx+z0MMry1n2bJluN1uDjvssB5Vyv369eOWW25hS1uCVxYs4J45s7n+7scIlPbi1FNP5Re/+AXd3d0sWLAg78WVlJTw+9//nnPPPZd58+aRSqVoaW6iuLwXJ078L+wOF6WV1cx6/M88f99MegVd0G8g5VV9uPD62zhoYE1+rdbW1ubDpQcddBBjxozh8ccfZ/z48QiCQF1dXZ6mWBT38pXnMGbMGI444giWL18OwHHHHcfatWt5+umnEQSB559/nj59+nDggQcyd+5cAKZMmcKYMWPwer3cc889TJs2jYULFyIIAoqicOyxx1JUVJS//tprr823rAaDQW688UYaGxs54YQTgFwefPz48T3W9V4D/NXq8GQyiWVZPTpxgLwnX1FRwWeffUZnZ2eP86FQKG8oLcviiiuu4NBDD+WSSy7JXzN8+HA++OADhg4dyvLlyxkxYkQPT97tdvPpp59y0UUX8eSTT1JaWsrpp5+Ox+PJh8IBYrHYN9YT/ND42Sj/H+PHkE3cn0TcXnxXq9FXzwdcNqoLXWxrT+YNc8gjY5km3cksiaxGmd9JwC3n869uu8TOriQfbOnKVzdXF+YIPL4qufhtc0AQSGQNdu0heXDKInFFpymaZWCJl5Z4mqSiY0omoqjTKeb6oFtiOVrJPkUeLMOguMBBc3eK9rjCuuY4deV+OpNZKgrchNwym1vjrN4VoSWaoTuZpTLgRtUNulMa0bSGUuCmOZbGIQl0JhQcNgndsHDJIk3hNEUeB7u6EtQW+xjdr4gNTTGiaRW7LHBwryAt3XGCARerd3ZTVehhY3OU/mV+Mi4bsaZNzL/7WgRBwLQsSo/9Na6jp9D29ydAENHdHsqPvYjdu+opEqFXwE1C0QmNnUTi/Sdo8vuxBIlwWmNTW4yWaJZin53RJ/2S9X/5I7aP1vHRRzq9JvwXTZqF07RIJFVcpoWomRxcFWR9cwwpUEbHu08T7HMA2YzBoHI/729sxrSgoTvFoWdcxOp597HasrA5XGgjJuHMtFHiczCyb4iP15lEkipWIovo8JNYt4RMuB19zTvUDjkE/M5cj7osYVgmkiBimDkRCPPAk/l8yXzWLXsDye4ga4AvUIiSjLL62VuxBJHDTjqb1oRK0ydv4y+uoP/QUXQlVQJuO+UBBy1RBQuBhKKh6jkDZQJpzSCpmlQEnNhEgV4hD5ddcwP333o9lmXh8Xi499579/sM7o1iHT7hFD77xzI+/NvrjDllEoufe5AFCxaQTCbzRVDDhg3jpptu4tlnn2XMmDGcffbZVFVV0auqer/fLe6zGXbKEgdWFvTYPM+ePZs777wTy7KQJIk5c+b0+HwgEOCss84CoLq6mq1bt/L888/vd6xBgwZx2GGHMWnSJFRVZejQoRxyyCHfuPYOPvhgJk6cyPXXX8/EiRO/FjbeF2effTbTp0/nV7/6FRUVFTgcuULOb/Pkv4q//e1vPYik9mJfT14QBO66664e50eNGpU/9u6777Jq1SpUVeXDDz8E4LrrrmPSpEncfPPNeU/+gQce+EZP/qyzzuLmm29m4cKFGIbR4zdft27dt1bf/5AQrH83efkTwH9Kk/lToNn8sSXj9p3jd431Tef7l3j4fHeM+o4kdptIPK1iIRBNq+gWFHtkKgvdCOSqsg3TYlt7gqxusLdnVAAOrCygrsJP1R42o+8qcNsdTvN5Y4St7XvyPQLopkUyq9G3yM07X3bglkWyShbVknHYBUb2LqQroeJ32bAJUORzsTucQjcsVjaESWR0XHaREdVBAi6Z4+pK2dSeZGtrnOZ4Lj8tSxK1JV6Wbu7AtKDU5yCt6lQVumnsTtGV1JBtAsOrA8TSGgG3hGUJJLMGadWgMuCiyCej6AZeh4xDEohkddrjCqZpklRy4hV15T52dqVIqgamYVLgsSMiYFoma3fHKHDLlPicfNkcpbbUT0XQRVdSwWMX8ThkKv0uHHaBt9e30pXScdtFBpUVsKs7CeQYwg6qDlBZ4GJ1QxhVM3HabTSG03TGs4zoU0g8raEYBhUFLkygM57FtCz6FnkIpzXce5SXWqJpXLINURDY0pFAQEQUclXqkihQW+ylJZah0G2nI6kQdMlopkW/Yi8+u8QXu2N7hEc8RNM6W9sT9C/10p3SSG1fSa+agRw0uD+rlvyF5p3bqPvFpXgcMs3RNOV+B8U+B1vbUsQVDZcsEvQ4KN0j9lEWcKJoJs3RDK2xLIPKfHgdNsoLnGxojlMZdOWr/gGGf086y93hNFvaEl87PrDMl3+GvwvftKasaPP/6bvnh3r3rVmzhnQ6zdixY9m1axcXX3wx7777Lps3b2bDhg0/uqjQZZddxuzZs/P5+x8D0WiU6dOn88QTT/xoY+yLnz3l/2P8O6xa/y6+K3/9bedH9A5SU+ylLZahoTtXLLWXmrA9qbI7mmVkn0IkMdcPnNYMFN3EtYeMIcfWZOQjAN9nM+K2SzkCCAFkWy7cLIvQO+TGbZcp8spYlkBWESn15wynxy5hDzgpK3DhtttQdY2M7qCxO0NFgQvNZ6JoJqIgIIoCWztTNEXSqCZEklkcso22WIaakJtxtUV82RpD1Q0CbhksC6/ThtNuI5XR2NGVpMBhxyXLdCQyiJKIXRKw20XCKY14VmNTcztjB5bx4dYORFGiI57hkD6FpLIaXSmVhnAaj1MmmlbZ1Z0mqxuM7V/EcXWlpFSdxq40w3qHiGQUNjZHESURye8kpSosXtvKLw4qx+tyYFoCAbeDpu4kRR4HbrtEdWGOurQrqeBz2HD7RKJpgwElHlRdRzUM/C4bPpeL3d1pOpMqGU3ngIoC4hkd04JIWkW2CdgkCY/DRndKpT2mIEkiBS4bkpirM3DZc3l+VTMYUlGAwybQEsuyamc3dknkwF5BNrfGMCwIeuyM6BPELubC/PXxUla9/DDrnE500+Lwc65kdL8QIgJ9Qq5cK5NuIUo5xi6XLBHL6NgEcs+cTaLIK+GwiZT7nRTu4U0X2UNoYxN7PGPfl87yh4hifdP63hz93l/xk0ZVVRXXXXcdjz76KLquc+uttwI9PfkfEzfeeCPPPfccN9544482xvPPP/+/5iXDz0b5J4H9MVX9kG1Se/Fd+evvakUCyGoGmrFvjhk0w0Q1TBTdwG237ZFJBJfLhrLHcudC2F/pXf6WDYJlWXTEs3TEMsQzOl0phepCN4UeG16nTKFLojroYeWubjrjWdpTBiP7FNIcSeN25Ji6bCKYkg0JlRK/g+6kgm4KeOwiPmdOQlGWBKJphR1dKfoW++hIZHHKNj7Y1oVmmAzvHSTksVHkdfLOxnZiGZ32eJbaUi+pjE51oUxnUiGpWGi6Sk2pj79vaMUh23DYBI4cXEZHPEt1kYe2aJaaEl8uTB1w0RzOYJckIkmVHV0pvA4bTllk9a4IZT4ngijk5BSjaYIuB0lFpzmcoSWSocTn5JDeAUp8Dj7bFUaURL5siXB4bQnbOxJUFzr5tCFKUtXRVBOPy4YggEOSQLA4Y3gVhqqhGBZb93rrlkmR14HdJlDmd7KjM0kyq+OUBQ7qFaQzqSAKArIkoJkWSUXPyVwaVo6n27I4uHeQ7kSWtriGppv4XXZkUURVVU47uAJNN0EQ2NIWw7DJFHhEHGX96XvOTGpLvOzujBIoDhFy24imDTTDwu+0gaVR6rMTy+R0oEUBErs3s3jJch568CFSqoGnWqIjnqUp32sOQyr9OYOoGt+4ju6//35qamq+FmrdXxHk7g2fcO+0P3DeeecxderU/PEXX3wxX+28b74U9r++v4pHHnmEN998k5KSEiDnnZ100klcfvnlLFq0iIKCAo455phv/Px/guuvv56Ojg6am5vztKYDBgzggAMO+Npcvori4uJ8nnlfNDU18c477/Bf//VfQI685KqrrmLx4sUA7N69m+nTp2NZFhUVFdx5553s2rWrR8j4iy++4A9/+AO1tbXcdNNNWJZFQUEBDzzwQL4V68orr/xRDTLAtGnTftTv/yp+Nso/QfxYIe3v2vl/VytSUyRDQtHZ3p5TaJJEAcPK0Rla1j95r20iVBV6sEm5vPBexZ2hvQJAztv5KhHJXuzdIOQIRzL43XaqgWKfHdOyqCsPkFI1TBNUzaR/iQ/BMilwO0mrKgXOHPvWkMqcClBa0VANky8aIxR5HRgZhVK/i3XNMWRRYmNLlGKfC5/DhoCJSxbAnaOzlESBdU1RBpX7sdtsxBWNXkE31SE3Qbcd0zSpCuXy7UlVp8jjYGtbHMUAm2RR5vfw6a4ISUXDLkkMKPXRlchiQU5XOJqmf4mP7pSK3Sbickj4HbY9nNhO/G6ZrKYxqNSP2yGzsztFodtOSjPoSql5r7F/qR/TMqgqdLGzK4kkSYiiRFssRVrVKA+46EqoeJwSgQI7a5uiNEcy1FUU4HVIdMYUCt02BpV6EESRtKojiKBoJoVumaDbSVcqSySl47SJHD24hI+3d6HoueK9AWU+dnTE8dglPt0VptTvZHt7kkP7FWGE0zhkCc2Cj3d00didodjn4KBeASJplY64gomZe7axqAk58blshHxORFGjKZKmVdH4uL4bmyRS4LRx9KASwEJpCaOIIoVeB8E9m1iPw0b/Ui+CQI+CwsJ/Y73sz8tdOvcTpk+fvt8ipvPOO+/fGOWfOP/885k0aRKQaxE66aSTmDhx4vfKy/4neOCBB4DcxmAvrSn8My/8r8KyLB555JE82cbrr7/OCy+8QDj8TynO++67j3PPPZdTTz2VBQsW8Nxzz3HFFVfkDfzbb79NSUkJRxxxBHPmzOHEE0/kV7/6FQ899BCvvvoqU6ZMYcqUKTzwwAPcfffd/8n0f3L42Sj/BPFDt0ntxXe1P33TeWHP+EBeoamxO0PIZyejmZR67fvklHNGdn8eCtDDuO/oSO7xfuW8YfbYpbzRru9IksxqdKc0ElmdYq+D9niGWEYn4MlxHgsS+BwSPqeMYVr4nDYGub2I5Dxtl0NENy08dhttiSwHVRYQTSlgCSQUne5uhR2dKYZVBRhY5qcpnOGTnWEymolhmtQWeynx2gmnFIZVFbJqRyduh4M1DWGGVAaIZ1SqQ166Eyq6aeYISxw2ygocdCSyZHQDWRBw2W18VN9FdciDbLNw2W2kVYNERsPtkHApIkGXHZsEzeEMkihidpsMLvfnDGJHEtMCw7LIqAY+p0yhJ/f7J7NZjhpUws7uNJta4qgGyCKYgkWx34kkCpiYtMVyEQ+vQ0YSBdoTWbJqLg3gd9lpDGdybWgCjJYKqQ656RV0sXJHmOqQh4ZwikQm5x2fNKSCWCZHKyrsUd3yOwRa4woeR65wafXOMHFFp6LASTyjUeR1EnLbyeoWH2ztpHfIQ0LROahXAeFklqQKbstEQEQSchSoxX4nq3Z243HK7OxKIkse1jZFGFtbTFjNFXTt3cTujuTkLr9cs4oVf3mBoM9NMBhkzpw5bNq0qQd1pcvl4vHHH6ewsBBN06ipqemxVuLxeF48wTAMrrnmGtLpNMuXL2fDhg0Eg0GGDRsG5AxyLBZj1qxZDB06lLVr13LhhRfmW4XOOeccVq1axUMPPYQkSVRVVTF58uRvXauRSCRPgLHXWNbU1PDUU08hyzJNTU15T7qhoYHp06djs9morKykubmZuXPn8pvf/IaGhgay2SxTp07l9NNP/5ffGd9nLnfccUcPWs8VK1ZQW1uL3Z57VxUUFHyt73f79u3ceeedQK5Cel8POZ1O88gjj+T7lQcPHkxbWxuQq9QuKysDoKamhh07dhCJRAgGg//y3H6q+Nko/wTxQ7dJ7cV35a+/6fy+RlpAoLLQTYHbTumequuvVl/3qOreZ/xwSu1h3ENeO43hNH5XAZIIlQFnPvyoGSYbW+I4ZRFVN/C7ZDK6hiwJ2CUR08qZcVkScdtEkoqOYZlEMypum8iGljgf74zgsUsoukHIbcfvkFE0i/akStArs709R5FpWhaxrEZrNMvqhjAWVq7IymtH0Q06EwomAqt3djO0V5DNbXHcdpmORJbKoIvORJYhvfwoBrRGM4S8doo9DjKqSTKhMKpvIWubomhG7p5tkoSmaRxg6+DLv77DgWddhUMSkCSB9miW3sUeUlmdWEZlw+4YRw8qpiposbktjsueK7aKpzWSWQ1ZstOw/FXeb+hNsmIEboeMpegoukla1Sny2BGFvQINAu2bPqN1+Xz8Bx5N0WEnUR10UuAt4J2/vEqycjTqjpUU6BE+Mk6gd5GLYp8Tl93G2t0RClx2VN1iW2eKSEZjWHWAbe0pkoqOtv5tujZ9jCb7iLhsqOkkWsVBFAw/meimFcQNO/aBIxhY5mNdUwzDNOlMKMSzGl+2Cgwu85NUs+iWRXWhG7tNQkDA77JR4nflcsSCRcBlRxJzldt+p42MLPHf117HzsZmOtrasASRVLSL2hFHMHLcYXyxYimPP/44Rx55JOlMliee/BMO0eKcM05h4cKFWHYP0666nJSi54VbFi1axOzZswmFQpSVlaHrOtdffz0rVqxgyZIlnHTSSXmDDLnQdUFBAfX19ZSVlWGz2XjmmWdobm7mkksuYeLEicycOZN58+YRCoX43e9+x7JlyxgyZEiP9bmXu7m1tZXS0lJGjhyZD/fuRUtLC3369GH+/PmMGzeOyy+/nHvvvZfLLruMI488kldeeYXm5maSySSrV6/mlVdeAXKGci+WLFnCXXfdxYUXXtgjBL9u3TrGjx+fD8HX1NR8r7m89tprTJw4MR+CVxQFh8PBqaee2iMEr6pqfqzBgwezbNkyzjjjDJYuXUom8893zKuvvsoJJ5xAYWHu7VFWVsYDDzzAm2++iaqqbNq0ibfeeosdO3agqipTpkzh5JNPRtf1Hp7+v4I33ngDp9PJ0UcfzYwZM2hubkZVVS6//HKOOeaY/MZHEAT69+/PbbfdhiiKPProo7z//vvYbDZmzJjB0KFD8985Z84c+vbty6RJk7Asi+nTp3P77bf3IH7ZH35m9PoJ4sdok9qLrzJvQc5Y7g6nCadyi+arzFxfHVdAwOeUGVTup66igEKvA1EUv5HRC3LeTFssV3SU2LPpqAy6qCnxUuyzM7x3kBK/M58P1A2D6kI3aVVHFCTqOxN0J3WWfNlBczRNOqvTr8RLZcAFlolDFuld6CGlGLTEFbqSKoZpkVJ04hmd5ngW2SbgddqwiSIuWcIkx4WcyGgU+xy0xtOU+l0kFYOmcJov2+K4ZBsOWUQQLGySRGM4jaKZyDYBURBw2SQ8dgmvQ6Y7nmZIrwDprIbXacPrEBneO4ii6VSHPFQGnZQHnOwOJ9kRzrGEyZJAVtE5qDrAqN6FHFZbRDiZZXc4g8suk9IM1jXH2NgWx+e0s60twcaWOBnN4IBeATa3xcC0cMo2OmIKKUXHbpNIKDq9Cz2EvHbSisHAMj8Dy7y0bFqN65AzGDDmJFKqxrbuNJ/titDy8WIGlvvpV+TBMEwEETKaidshkauxFtCMXCSiT8hNTbEXEYFoRiWl5v6e/qETOOeGuxl/6e2Mu3QO7FpFgZBh+LjjKRs0Ar8rV3wVdMsEXHYcNpESn5NMVqfImyNiGV4dZEiFjzK/E4ucvKNu5KhQKwNuiv1Ogm6Z3kVuKoMuAG767Z1MvOFeQgeOpeSgo3CFyhk3+b8JpxQKi0rYtm0bu8MZfMWVbGlL8OGGndhdHjaHDT7dFaG070B2dqb4uL6bcFLJUzc+/PDDzJ07l/nz51NUVET97tacZGlW/1bGvbq6OgRBoLi4mGw2SzgcpqOjg2nTpjFlyhRWrFjxtZ5byIWvX3zxRR5++GG6uroIBAJfu2bAgAE89thjuN3u/Mu9vr4+v0kYMWIEkCPNmDFjBjNnzuTaa6/tYRCXLVvGBRdc0MMgQ44h69+ZS3Nzc485jBs3jptuuomFCxeycOFCuru7OfPMM/OeM8DNN9/MsmXL8tSf+3q6ixcv5pe//GX+//feey933303f/3rX7nlllvwer3MnTuXcePGMWbMGC644ILvJZn5TUin0/zlL3/huOOO44033iAQCDBv3jyefvrpvDd/9913M23aNObNm4dlWSxdupSNGzeyatUqFixYwIMPPpgnOgmHw1x88cUsW7YsP4YgCJxyyinf2l62Fz97yj9BfFeY+YfC981d/6f3kw8vhjM0RzJY5PRrK4O5XG6folzv8u5wOv8Zu82GLAoM7RVgTUOUsgIX7bEsXoeLXeEM/UpzbS9VhW4SsRiG3U1TOIVpQWdCRTNMSrxDYfahAAAgAElEQVR2WmI5TeWgIOO129jUHEOzTEzLotzvJJZR6VvowilLuGQX72/pRCC3AbLbRCIZlV5BP91JlYoCB5WFbhJZHdO02NWVpD2hUOzz0xZLUxbwkFY1jqkrpak7zcFVQTa3xxFFkayq0L/EzxeNYcoL3OzsTlCDQFLRUS1oiWapLfESTatozVuQPnuDdsGG1+fHe8ZltOyqZ917L+N1OfANHofN5+G9hW9gOTwIpsHuaAX2CpG0YuB32YlEIrS+/xLdloai6Qw57Vd0x1OI7ZuxEi00FQSorK1DUQ3iX7yNlkmycuGTyMV9ybbWY+t6gpiepj55MkNGTqBj50Ya/vEasixh8xWTGHUOBW4HQbcDSTRRXTaykkxXUqUpkiYVC2OZBhUhPx8tfhFPQSEZbzE71/4dzRJwKFEcBx1O/yPOINbVwty770e22SgIBHlNSbLgzy/y0iN309DYiKoo1Bx2IiPGTWDHR28RLKsgW3oYWc0gklbZFcnQ0JXEskCyO9GVDNsamqkQRdas+QzT0Nl0z524vV4ELLZtr6exYRc3X3kRwdJeGNkkVQf6+Ki+m/aEQlssSyAQ4NNPP6Wuro62tja6I1E2tCRZvXoVGzZ+ybPPPM1dt9+a50Le10irqsqll15KPB6ns7OTzZs34/P56N27N7Nnz+amm25i3bp1QM5Da2lpob6+ntdee4033niDkpISLr74Yh599FH69+/Pzp070XWdyZMnIwgCY8aMYcWKFcRiMW699Vai0SiTJ0/mueeeY+3atQB0dHSwZs0adF1HVVVuvfVWSktLURSF5cuX89lnnzF06NAeIXhFUXjzzTc566yzWLt2Le+99x6RSITq6mpOP/10FEXhyiuvxDRNSktLufvuu1m+fDlud8/2sMLCQhKJxNdC8JlMhpUrV/LUU0/R1dVFJpPh9NNPR5IkBg0alPdwGxoauOmmm/L5Zb/fn2fcKikpIR6P58dKp9M9SD2WLl3K3/72N6LRKNdccw3jx4/n7bff5vnnn0cURUaMGPG1wrXFixczZswYAE444QSOP/74/N90L4HIxo0b81zdRxxxBCtWrKBv376MHTsWQRCoqKjAMAzC4TCpVIqrr746T96yF4cffjj/8z//wxVXXIEofrM//LNR/gnif6tN6vvmrv/T+9k7jk0kL4PYlVQpcMlUFbpJZXNUhW77Pxl0bCJUBt00R1J4nTaSWZ1CjwPdtPA5bIhCjhwikdXZHcsQ1XQiaZVCj4xh5XpW/S4Zl13igAI/AXcuFJ3RDYZVB2mJ5PipB5d4ie1pPbIAURBRdA3DBFGAoNtBLKOh6zoH9grw4dYO/C47HfEMI/qEsCyDYq/Mjm6NJV+2AQJOWWR4dZCADerK/JT4nNgkkTUNXfhcdhCgttRP+7YG7DYJr93Gzq4UiZRCdZGbxMd/ZtiUW7B5C9nx0Vusens+VlkdomVQO/E3hBMZts2/lWOuvofmlMDWhQ+CojOkws/uSJqQRyb2ybv4eh+IfdBRSGqcv7/4Pxx73cNUDhpG/+FjqT7gEDa3x4lndUpGnULz6iUMPPECGj57H5ss0+f0axka1HnvmbspHjqOxD/mMeD0G3H5AnSv/Av++HpK+hyJYVhUBty8s0IhtmU5zetXoCbCuP2FDPnFr9ke0fE5ZeyyiMtjJ5ro5uybH+aNNbv48rXbcAw9kd1/m8fgI89g4LBDqf/orzSuX8WWpi62bFjLidfcg26YtG9bi9dhZ/xpE9nZkWR3JEt3OMPGz1dxyxXnYwkQ7e6i1/BjqDvl12x49XdsUJPo2TTPzFvIX9/9gD8/fj+aYbHo6d8x8uwr2PbhYlq2b0Cw2RkwXMQmgqEZdCcVmlvbePTRP/Dwww8jiCJTrpnJu6/PoyAY4qRzLqC4rIKZt81iwfyXgRwDWH19PWPGjOHzzz/n1FNP5dxzz+W4445j5syZzJo1i2uvvZbt27eza9cu/H4/uq6zbNkyrr76at544w0OP/xwZs+ezeuvv8748eP54x//SDKZ5IwzziCZTLJ06VJ69erVY20NHz6ciy++mIsuuoizzz6bIUOGYLPZKC4uZvny5fkc7F7pyHPOOYcBAwYwYcKEHiH4yy+/nCeffJJTTjklt/5sNi688ELWrVvHn/70J0477TQikQhvvPEGmzZt4pZbbuHEE0+kd+/ePchXnn/+eex2O83NzQwaNIjZs2f3ULSCXAh+9uzZ3HHHHfz+97/n7LPPpquri8suu4xgMMjMmTN7XD9z5kzuuOMOTNPEsqx82xVAY2NjDyKU0tJS7rrrLlauXMnTTz/N8OHDeeSRR1i4cCEul4sbb7yRFStW5I0w5AQ89hbTeTweIJe7/u///u985fW+mvT7cmjvG83Ye7x3795UVVV9zShLkkRhYSFbt279VmGLn43yTxTfp43iP8W/krv+T+5n7zj/1GAuIKvrBFx2Ymk1L6FYGXDSK5ALYe+9trqwiLKuJBnNpLE7jSiJuGURp02iK6FgFwRK/R46OrIksjq6blJakOtZTmYMwikVp02kOuimPZahOuTh4x1dJLIGkggBV5AtbQmKvHaK/U62dyaoKfbikASCHjuprEI8I+G02/hgSzuVBW6cdpGAW6Y7qTC43EeRz8HHOyP0LfKytT1JQlH5uL6TgWNq2N0YpjWuEEkr1BZ7GSgIOG0y0YxCxCYR1w3WN0cp9TkYXFFAc3snluQgVFzG1vY4/qqBJBo24PMMI1ZQSlXIhdtM0iA5Wd+hMqi8gPaq/uCQ2dgSZUBpAU5ZwIy34znwcLqyGmX+IJLdhaClMC2LcFpF7UjQFsuiahZ9QrkwsMMmMajMT1QYQLHPieS0kc5k2dnaTSoWoflvf0QQQFMUFP0AMhVp+pX6+LI5hiQI9BtzMts8QxGjTWRXvkCorBeK7iAtiSiaSTyu4Aj1YmtnmoDfjyLnKuojbbup7F+X2xSFemOxks6MyajTL2LF/D+gKxn6DBvH501RxtSEiGdNOpNZ4vZejLnuj5T6HLjsIqvf+jOyJ8DBI0cxctShiI2r0aJtVJaVcNCh4/jL3D8SDneTjkfYtfpd7E4XuiUQ6HMgLcGhaB1JupMq7ZEMB4wez7kXXUmvoAu3Q2JrW5L3//oqpmmyZNFLACipWP4Znzt3LlOmTOHYY49lzZo1HHLIIaR0gRdfe5vLpp7DkCFDOPLII7n44ot5/PHHcTgcrF69mtbWVvr168czzzzDk08+yXnnnUdNTQ0TJkzgtNNOIxQKMXnyZOrr6/nyyy956KGH8gZl2LBhjB49mlWrVvGb3/yGl156iSOPPJI1a9YgCAJlZWVMmzaNuro6AN5//31KSkq+0UvzeDxMmjSJRYsWUVdXx1lnncUpp5zCiSeeSCqVwuFw5I1USUkJhx9++Nf6d88//3zOOecczj77bDo7O3vwVe9tXRowYADDhw/n9ddfZ8yYMdx5552ccMIJDBs2DL/fz4MPPphXtwKora3lhRde+Nr9xuNxKisrexj9Aw44APgn73djYyPhcDhPvZlKpWhsbOxhlL/Kod3a2sqVV17J5MmT83KY+/5m35dDe38oKSkhGv32JvWfjfL/w/gxc9ff9H17NZhNSyCrmUjiP73t5miW4dUBSr4in1jgkmnoTpHVTBKKRqnfSUcsgyCKNCayKNkspT4HWBaF3lze1eO0M6LaS0rVMfbkJU1BYHNLfE+7k4hhmTTFMgwo8+Gy29A0jSMHlLCmIYLhkNmyo5tD+4ZQNY0Sr51inwsQaOjOEMtoFLhkhlYG+GBbFy5ZoimS4cBKPzs7EwwuL2DFji4sC7IpDUmArqRGRjFpySYJeOzEMiqWBWUFTnxOmY3NcSSbE8FQ0BIRTEuiY/tGRF8pnfEMmDlxDtnlQ9AyFEgq9R0J9M4G3DWFVBT72NgcpdDnQHMXk23dzvAxAxGzMTakEmwLmyQVgxJBIJHV0A2LjlQWWi0QLEzTJJJWiGdUnBmVsC0Xkg0FA+AqYNSvbmBzt0bX1jW0I+O2oL49wfDeQSL2nMdfHnCg+voRiU/gkz//Dudx0/Dv+RMbe6rH06pOedBFO+Bz2LAFK9j25XpKBw6nraEeRTNZv62Zz9etp+ToS9FUlVV/upEDLxmOZpp0pbIoukVaNXDKIru6U4yqCZFSDEoLRNyyjdH9CmlIutkZEwi6ZSoCTkzLQnR4cRUUMuW6O9gZM2je9Ck7oxrCnlSCLAl0pVSK9jyXTZEM/UtzL/2yyt5UHzmQkUccSyIaYeenS/f7vNfU1PDWeysYNr6IaHcn3ZEoHVmJCRMmcN9993HMMccgCAIPPfQQhx9+OADz58/n6quvJhQKceutt7JkyRKA74xGbdiwgfLycm688UZM00RV1Xwlc79+/fIh+Pb2duLxOCNGjOjB8b0v9g3Bf3XcYDBIWVkZjz32GD6fj6VLl34tbL0Xoihy0003MX/+fK677jpefvnlHuf3N6cBAwbw+eefc+SRR+ZD8N+FHTt2cMUVV3zrd/fq1Yvy8nKeffZZZFlm0aJFX2My2xtuB+jq6uLCCy/k1ltv5bDDDstfU1dXx8qVKzn00ENZvnw5o0ePprq6mvvuu4+LLrqItrY2TNPMF6d9E74Ph/bPRvn/Yfxv5a73N06J105W60lCApBSjR5FaJZlYWLRGEkTSSvYxByzV4FTxu8SiWU0EiYUeu17VKhgUIWfRFanNZ6lyGMnntUwTIuAW6bAZSOlGDhkgaqgF9WwiKRUklkdQQBV16ir8KPoBhUFTiJJhWHVAVTLIqPqdKoGyaxG7yIPPqeNuKISdDv4sjmG225jS1uMMbVFOGwSm9sSOOUcrWV5gYvNLTGG9ykkklHRdIMir5PG1s00vnIXorCnknzc+ShDz+L9P92LTZIQ7C4GnXoJyc4mOlpy1JGKZjL2l5ew4pX7MWUXPreT1mgGIaVSXeShyOtAGXIc3StfZtmGT9BUFc/oSRQVeGgTwGTPhkgS6B30kNV0XKFKOpY9S7KgL+m0xiC3gy2tMQwzx6Y16MT/4rN596HpBpLNQfHYqcSyGg5bbjOS1U06kgojR4ZY0xjGefBRdLespbz7c7o0A7tdwOuQUOw2BAR2dCQxLYsdHUkGHz+ZHX97lk0fvIHsdOKzSYQNB2TjbHl5NoIgUjTieMqCHv7x1qtI/hKM0rp8d3uRz4GhGxR57RR57VSHnAgCOVWpPc9Qdk8xGoLAyVOu5M8PzUTTDWSnm5rjL0Z25vrDdROcNhFBz6lR6SZ0J7ME3DLHnX0e8/5wLyv+vhhDzXDDtdf0eHYtyyKW0TjmjF8x547beG/pEjRVZdLlN9Ka0Dh41BhmzJjBbbfdRjwe57777st7hEOHDuXSSy/F4/Hgdrs56qij8i1B34bXXnuNWCxG//79uffee3sUTF166aXMmDGDd955h2w2yx133EF19f65uCFnxG+44Yb8RmFfiKLILbfcwiWXXPKdnOEAo0ePZvTo0Vx44YX8+c9//s553HDDDcyYMYNnn30Wn8+XV4q66667OPPMM/dLCXrwwQfvtxhuXxQWFnL++eczZcoUDMOgsrKSE088scc1hx56KGvXrmXkyJE88cQTxONxHnvsMR577DEAnnrqKW6++WZmzpzJgw8+SE1NDccffzySJHHIIYdwzjnnYJpmj7D6/mCaJu3t7dTW1n7rdd+L+/q9997D4XD0+GO9++67TJgw4bs++qPi/w/c1z80vsoE1t5Y/61z/DGYw77POJZl8Xnj18M4X+UlDqdUPtjaSX1HMn8sJ4LhIOSxs7k1SVNnN9WlhcSzKl67zNaOBKIoIgkWh/crIuCys7M7RdBtZ2t7HFEQ0U0Tn1NmW0eCweV+dnamiGY0upMKB1T40Q1ojaURRRER6Exm6VvkJZJWSCkmaVVj/KASfA6Zdc0xsprF1vYEQ3r5cdlFyvxOtrelUE0Tu00imlZQdJMBJT7SWq53uzWWpqbQTXdap1ehm1hGpTWWpTWaRbaJeO0SfqeMQ5aIZlSqCz2IWISTKqplklJMVN2gd6E7V6VsWrhkiXhapcjvYGNrAizoTCoUeR1IAhxQWUBXMktXXKM1nqG8wMXucJohlT4O6BXgH9u6CLrttMUytMYUgm6Jk4dW8tb6FoZUBljfHKPY52BHZ4raEi/N0TSH1oRojWTwOGQyqsbY/sU0RzNgCaxpDNO/1IdqWNQUeXBIFl+2pdjZlcJjl7AAdcdqBtUdiFhQgrntH7Tu2kqf4y7C5bCxeQ/3tN8hMbZ/MZqRa6XyOmwoWq5KPJxS6FfkJa0aaKZFYyRDyGOnpsjDyL5B0qrOa5+1kFR1vE4bxR47FQE3WU3H57azoz1BWv//2HuzWEvPs97z983jmvc81lzl8px4ICQBknRohDjdcBNQWomUu0bcEKmFuADJIARIgEAtlNwgAUFCnESoTwN9GpAPnJwQOpMdD2W75mHPe695ffP49sW393ZVucouO7GJk/pLVml7r+Fba7/f+7zP8/yf/18QJTmWoWBrCg8vNhhHOWuDkGMzLvX9Eb67uZzdTJoM05yLu/4hmfHArvRAM1sIwQuvXmJqful7uuc+85nP8Mwzz7zBXele8IO29/3d3/0djz76KKurq3zlK1/h+eef5/d+7/cOvZRXV1fftff2fZ9f+ZVf4S//8i/ftfcA+OpXv8orr7zyhuz+drxlpvzMM8/geR55nvMXf/EX/Omf/im6rvOlL33pPzwovxPcHBziUrmlgf9+x53Y1ISCM2/yGd+L3vWd3kcIcU9ZepDkJFlxi/513awec7kbULcUjs7W+c5G1Ze9sOvTcQxMVWapbbMxiBlaKXuTlN4kqcauhgHTdZskK3hgrsHAixhGOXVDRRYw5Ri8uutxeq7O1y93aTsmPT/D1GKOTztVRqtLaKrMczdG9MOMoih4bLmJIoOMTFFI6KrMufUJiizxU6enmYQp+f5oWMPSaJg6c02LTW/EC+sjkqzE1GSOdly2xyFRXvLYrMskyAiSgu9cH1CWgh873gEEl/cCHl1s4BgKm5MYUQhcQ2Gl0wAh6E4SkCUalso4yug4JsMgpW6oTLQcXZOZxBkPLzeYBAllIUgLwZWej6kq1EyVlbbDa9tjFlo2fpKx0rG53gt44kiLIMl4+ugUfpxhayotW0VVJC53fY50bP7uxS2mHIMwLyjykn88t83TR6dYG4S0Ha2y9FRkurUWz3/l/6SUNWxd5eFf+N9Ji5I0qDS7CwTLbYfVtkGQVaN013shPT+hF2Q0bZVCVONT68MQISRURWJtGO2bgqgUoip1X+0GXNr1WW5H/NjRDg1DYaXjsjYIWWqayIBtVOtrfRBQs1TirMqyS9I3uJwd7Cc744j1QUVmNFQFCSoyo61Tu0kx7+AefXFzwpLkHa7775cBzfsV8/PzfP7zn8eyLGRZPizBf+ITn2BhYeFdfW/Xdfn5n/95/umf/umQef39hhCCv//7v+e3f/u33/KxbxmUL168yF//9V8DFZnhV3/1V/nCF77wpnN6P6i4PWhtbE6w2t4PzQ1xRzb1IGQYZu960H27uFdGt6MrSEIQ5wWXuz4SAgE8utSgZqooMlzaiVAlmbwQSMh4cUa9ZXNxt9Junm8aIEmMggwvzVhs2hRliaUrXNrzsFQVP86wVIm5tsXFrs/WKCLPC87MN9kdR9iGQlEK+mHGYtMgyeE7N/qEaWW8cWq2VpU/C0FWFgy7E+qmxtEpmzAVDIOEswsN/seFPU7NuPhpQdsx2JrElIXAVGX6fkLPL5HacHymhp9krA8C+l7KrpeSFyWSBF6U06np6KpMLuD8jo8iga4pPLc2xNbVyqAjK+j5CaamUjc1BAUCiZZjYOkqCIGmKnRslY6jc3XPZ7FpcmWvACFYbpkcmbLZGSdossDSFbw456GFBssti0mS0/OrCsBS0yLJCtKsUl4ripKnj07RCxJsVaaXVixsSRIUZUnN0tgaRCiqTN46xkOfeQZVVWgaErKisucl1CwVNQdEga1XlpylkJhv6GiKxLFplyeP6GyNY+KsQJIqrkLdVPbniQuirBpfK5F4eWNMXgrqpkaWC8Ik56ljHU4CTx1rV9Kcuoqgsqu0DZVJlDOJqjL4lKtzei675WB5sJ/0/ITNYcRK26bjqIdTBklWUDPUwwPnwT168+75TtX67qQ5/X7Fk08+eUdZz3c7IB/gF37hF97V15ckiT/8wz+8p8e+ZVA+mHPTdZ3PfOYzh3T29yNuD1qC74985Q8K7sSmFtzKpn6vytX3grfK0g9MKRRZ4vyOR5jkWHpF/rqyFyAkUGWJcZSjqSqqLCEh0BSVq3setqEhSeAnBVf2PE7N1rjSCxnHOQ/O19gYhFiqiqJI2IZC09bYGcU4ukLH0Zlv2Fzam1RBXIh9He+cjt3g3PYEQ1WIksqvemcc0XEM8rKkYesM/AJTUzkxZbE9yZhtWCiSxAMLDaKsYGsc0Q8yZEnC0GS8IKXjGMR5SSkgywsWWxZBnPPyeIKtq2iKTJhkTOKcOC+I0gIZgWOqKJLEua0JRQmmBuuDiJajsdpx6PoJJYKzCy1eWBvycjhmrm6w2LLICzgxXSNMMjJHw9RkXE1B12XSvCSIM17dGvHhE9Nsj0N2vJQTMw6jJGd7GNCwDF5cH3OtG3BsymGuabE7CphrtLncHaJKCntZRNM2cI3KfnOl7XB51+P4dI28KJh2TaZcnV6QYmgqr217zDVMHl9qoqkKaZ6zPU54cWNCVggcXabjGqiyTKdmcmquhp/kjIKETqSRFiVhktO0VGxNpRRl9bfWVSZRhpdkTDlN6naldnYnG0Y/yQmS161HAfp+Rcw7wM37iaFVpfjNYYhr1jE1ibMLdVY7NnMN6/A+e7fU+u7jhwdvqej12c9+lp/7uZ87FBP/tV/7NeI45rnnnnvXL+77jTe7IX4YcCfWtHTT/z842T9/Y3jLv/9RVQ8hxC1qYrdfR2VKEWNoMg8u1Dk1V6PlaLQcnbQo6Vg6EhJ5WTCJMsZRRs3UGIYJpq7S82Najs7VvYpwlWQFU45Ow9BwdI25mompChRJ4vGlJpIkMYoytsdRxeamZKXt0HZ0VpoG8w2DRxYbRFnO5jBgEme4lsZc3WCpZTNb11loWuyOYq4PIs5tjmnYFqoiuLLn860bA7yk4N8v99FVFXPflnJ7FGHs2wuaWjVutdi2qWkKTbtSwZIAQ5M4PlMjK3I6jsbZhQarHQdFqpjDoyhjFKWHWWNRCGbqJnVLR5Elel7Vq94YhKS5oBCCmimzPYl5bn3MN68OeGF9gmmqnNsYc2HHJy0EHzk5zc4k5Ein6q22LJ1vX+3jJyVXugEzdQNTV2g4Gjd6Hksdl//r+U02hzE3hj5N28CPMk7O1chLsHSZnzg5zVzD4LHlJjVL5b+e2+ZKL+Cb1/ucXqizO45QFRlbVw4tJFu2hqVJpIXg0p6PrkokWYGERNOQaTuVv3LXi0nykvVhxDhK2fNiOq5Bw1SoWRrHp1zaroZrqnedNDiwfTwIydL+zzdvmDfvG66hMFPTyUrBd9eGXNgN2JvESHDLwfe9mni4j/cvlGduHgi7A06dOsWnPvUpXNdFkiQkSeLjH/84H/vYx5ienubZZ599g5j7e4Ver8f09PQ9P74oBdvj+PDnyWRCo15tbNZNwhXvV5iaTFaUTOLXN4u6mnN2dRZJkhiG2RtM2ydxTsvR3/PPf3BAuLDj0fcrglNWlHTc1yU6e35C30+RJZm+l+DFOWuDiCQr6YcpJ2bcyvDe0tmZxEzXTVRF8PEzs2wMAxzDwI9StiYJiKrEvD6M2JkkDIKUmqVi6QpZLg57gXFecrRjUwqJ87s+fT9hoWlxar7OsSmbi12fuq7ipwUNUyXOBD0/IS1KpmsGl7o+UZajKwrLbZudScRszarGqMKMshQ0HY3rvYDlto0mSzRsDdvQuLLno6sKdUsjiDJGScpy0+L4TKVepsoSyy2DpZbD+jDg3OaYPT/hyJSDKAW7kxhLVWi7OkgSuiJhqDLb45iBnyHLErIEDVvjIyc6uIZK10t5ecujHyS03EooZWMQstRySAsBEgRRRs0ysA2F5Y5DTZdZH8Yoisz6MMTWNQxF4ux8A12VWBuGlEJCCIGhquR5zuq0y0Y/4mrPZxRmhFnB9ijCNDS6k5hSSFzvB0RpwcYg4hNn56gbMoMw5xvX+myPEtYGAZqiMApTFEXG0VTmGiaWKpEWElf3PNqOzmLTqvTKRYksyexMYnRF4iMnpjFUmYatcXymxvFptzJcuUOlqCgFG6MQV1epWRqrHRvHUFiZcg/vlZv3EwkJW1dYG0QstizmmxYdV8eLi1vur4N7dHOvT73eAKqe8t2u493C29077+O9w1sGZajUXW5fMAd/0GeeeeZdr8ffDW93Yd0etLzJhLOrc+/5DfFuQZIqw/mWo1O3qtKlHA0Pv6ODIHc76pZGw3qdYHVQ4u76MaMww4syClF9f2/3ezrIhtf6IXtejCJJmJpyTweEg01PkUBVJF7aHOFFOXWrMsHo+glTNRMpizEsg5atoyoKUZzRDTK6XowsS1iawvHpijk9CjOSvCTJS17bntBxDBxTQVUV4jSnFBK7k4ir/QBNVXhosUlvFLPjJYRZyUsbHkICRa6yuF0vYaZm8OSRNmkh2JskdBydzVHEiZkaWV6iaTKvbU+IskqRaLFpkxWC2bpJmGW0HJ0HZl1qdlV6vd7zCdOC49MV12HaNTBUCVNTmXJ1nl8bcakb4qcFuiqzM4p5bKWBqWmkeUGYVmxuISDJYRylNO1Khez0XA1dUSiBc5sTekFCkBYokkReCgxVQZZlSgGbo5BxmHNs2iXaZ4y/vD5iqWmiKSp5UWDuu13N1Ey6XkzHrfSY14chWSHw4nfoLvYAACAASURBVAzH1BGlwNQrItj1fkiYFjy83CRKc270QxQZvKSkLAV1W+dYx6FTM/n6lR5ZLhhEKXkhsT2JONKpVJceW2ny4EKdpIRvXu1xrR+xPowxNYWXNiphj6atAdVc+oMLLkema5yZq/PEaoultn3H9SyEYK0fsD1OWB9GTPa1rs/M11m+6Tm37ydxVqApMosta5/sVT2uZqkUZXX/FaISySEJWJmfZrVtYxsq/aDSan8n99g7wf2g/IOL77lm8n4ifN1OLuqI+g8NyesAt/dpd2/63b2Uzm62wdscRvT99NBecbFlv63vSwjBhe0JL29OWBtUMpbTrs5TR9u4d7mWm3trLVtjsWlyfsej7ydMOQYrbZuyECRFSV4ILE1G0mSMUublzQlpXtkt9ryYBxYapFlREcAk6Lgml/YqZ6iun2BoCoMoJSlKLu8FzDVM8qzgwcUGo8sDLEVmHKY8sNzgO9cHOLpMmudsjmJmHI3TC02KElqOzv/z0jZLbZsozTk+7bDQzFAV0LSKwFUzVUZRTpxVpKWltsVK26IXKCw3K2bzi+sj/KTA0hTarsm/X+nx8FITL6lmpqUoZWtcEKQFErDatknzkr0w4dVtnwfn69i6TJQWnJp12RhWJL/HlxtsDWJqHZedSUwQF6yPIpq2xorj0PWGtJ3qMCeAvpcw3zDxkowwydEUSIuS17Ym+GmOrKh4kcdc3aKeF4emHkFScHxGZm+SsNJxuLLrV6QyTeLotMvFnTFt12S6ZiLKEl2W2Q5iJlHGkSkXU82xtGoWe2MUseNFXNjxWWlbzNQMbvRDFhpVBnpqts5MTUOWZbaGVak7SiszjlGUMlPTKASMwxxZlsjKEkVVeOJI6xbDlDtxLA7aJgfqc0leYKgKM3XzlrV/+35SimosTrqpDy0QDPyEizuvj/QttSxMuWCpZb0rvun38f7G9+wS9X5bPDe7JJly8b67/jvhrXqzBzgQ8bgZt48hHZBX/KSg56cIYG0QkpcVKW64L4l5LxiGGTcG0WFABuj6Ked3fcq7POf2A0KSFQgh0FSZQgjyQuDuZ/ZPrLb48RNTnJl3mUQ5EhINU2O2rvHEkQ41U6XvpVzuVsQhJKosORPI++VdPy7IyyrzTfMC3VC41g8ZhAmyInN+x+frl3psDCOirLIVzIsCy9A4v+cTpyVfu9TDNXW6Xsxc02IQpjy+1CJKc55caXGtG1A3NWZdnQ8e6XCt73FmpkbXj+l7KTf6Pld6ASttF1WuesqXdiccn6lTFCVCwMubY47NNDjatpipGzRtjd1JzCBMMdQqu9rzEl7ZnhDlJf/9QpdekPPC+phBmO97X+cEcYmqyEjA7rhSJltq2WRFga5ILDRMTs65eFHKXE3nQyem2BrFhGlOL0iZrZn826U9FFWl6yf0vARXV5mpmZWwSZpydrGBLMFS28LWZT642uZGd8IgygmTgjir5okVWcLRZH7sxBQ1U6FuKYfti/VBgKGqlEKwOYqpmxqrHRchVezp3XHEICgohUDXFIqipO0YXOv5nNuc4FoaRVlVWAxN4eRMrbLO3G/BHayvO3Es/KRa45X6nIStqyiyRJgWb1ivN+8nK22b5datpLGWrb+hOrUxjIhL5a7a82/nHruPHz68p+yCsix55plnuHDhArqu8zu/8zu3DIV/+ctf5m/+5m9QVZVf/uVf5mMf+9h7eXnvS9yr0xPc2xjSAXklSauTf1aUaEqVIVq69rZYokGSk+RvVO1K9olIB3PKQpT7M50aXpTStKp2yfNrI756oUucF3S9mIWGxcYowtAqVu3J2aoUubOzy1LLomEqCEnia5f7FAIcXeXsYp0zpYuMYMrWObtQp+clDIKckzNV5pgXBZYmODFTYxgkGKrMf3pknm/fGNJxdXRVYq5h8p3rA07P1TkxU2MSV+phS22H726M9ku4BnWjKg2PogxZVri85/HJs3P0goSVlsX1fsgji20u7HlsDmOGYcLDC419ApbPQtNksWnRsHSGfsRcw+FrF/douQb/7bVdPnSswyNzDmVZYqoytq6yOuXQ3h9rQoJ/Ob+LrWsISeLxlTZJlnNq1uXrVwaVDSVwcsZlEkPL1UiykuMzdtVr91M2R2EVZFoWXpwTpJU/c5xV/dE9L0FXZWZqBhd3J9wYhGiyhKGrvLjpo6kSDUPj7EKd17bGfOtaj8dX24yuD7kxCJhyNTqOztogIMlLRnHOcsviiZU2hgr/fmXA8pTDlT2PJ1fbvLgxBCT8KOOR5SayBIWQCNMcR1eoGTJzDZPXdj0WmxZRVtA0qpL5StsFCRqWiqMrt6zfuwXFA1nN2/FWZKw73V9BnB1qu9+MMCvfMRP7B2mC4j6+/3hPy9fPPvssaZryn//zf+aFF17g93//9/niF78IQLfb5a/+6q/427/9W5Ik4dOf/jQf/vCHb/HgvI834l6dng7wVmNIjqEiEIyijLX+gXMSHO842Lq4Z5aoEJWWcpYWFGWJY1SEqkJU4yOuobHcspmu6fz75T6DIGVnHPPalsdjy00eW2lwZc9HUMlP5gVsjSMeWqhj7WdmZ+brSJJEUwddlmk7Js+e36UQUhWEXZ2Lux4fOj7F82sDojhnZcrh+JRDCaz3A6Zcg41BwMMrbS7ujjA0jX4/YbnlMAwzdFXGVDQ6rkmQFOx4MZIkaDkGXb+Sy3xooUaQlbRslYapkZaC1bbDt6/12EoyHllqkiQFflpwZV9e0tZV5hoG0zWDNC+52vP44JEpel4CwI2+zyNLLb51rcdUzSJKc450bF7cGPLB1TaLzYKmVZWOESUbw4juJGW1Y3Gk47A9jlnrB7QdnWu9gOWmTcvR0BQZRZJRZAlVlonTAktXeWl9jK4p1AwJXVW40guYrVdBeaZu4Ooa7K8LTZG43gvx44xTsy7jOGPGNXlubcBy2yFJc7wk5+L2hAcXGtXnVSSOtG2SQtBxKuUvTVFYbGkkWQEIvDRnsWGz0KyCWiFgGCU8ebTDsY7DQ4sNZmsGSSH215Cyz0yX2fMSgrg6AJ6ccWnZCjtjSAuBJFXyl1vDiA+uvq5NfLegKMM7lp+9VzEeW5PfERP77RzC7+P9iXsuX3/lK1+55ecD147Pfe5z9/xmzz33HB/96EeBSrf03Llzh7976aWXePzxx9F1nVqtxsrKCufPn7/n1/5Rxfd7zKtla7RsnSDJqZkqmkLVnxyFuPsGEW+Fg17yt68NeHXbY60f8u3rQ/KyZLamc2bWPTzd+0nBxd3KBaoQ1Vz1i+vDai55P1tXJcG0a5AXECQFNVNltq4TpDmDoCoNNmyNYZQSpCVZVmJpKmUJpq5yedfnajfEMFRe2ZpwtReQZSUtS6dlaZyaq/P8tQE106AUgICdSUTL0tgcxsRFpZ18bMrhofkax2dqvLA+oO8nvLgxYqllk2YZTdPg+jBkz0v4Ly9soKoyHzlekWk+cnoaQ4EHF+sc6bic35nw/I0Rl/Y8DL1SlRp4Masdm6Io+V8eW0IIwULTYWMYYukKG4OISVzw3I0hkzijHyTsjEN6QUrd0JHkSpjjwYUGp2drnJqroykS0zWNQhQ8daRDx9HRFOi4BooscWkvYGtcjXCt90NsXWcYZmS5IC9LVtsWSw2bV7bGrA1jhmHKYytthkFCWgg2xyGnpuscm7Y5OuWC2Pei1mTGccE4SlEViX6Us+PFDPzKr9g1DK70fLbHMdcHIaWQGfkp13oeiy2boZ+QF4IkL2lZGqttkzDJGcc5HbdSyZKQKCWJEsGp2RorHZu5/Sw1FxKurRBnGVujhLVBhECivCmJuGtQNDVOz9X4wGrrln/fSdC7W8vIlIt7aifdjvsl7x9+vOUO+w//8A/8y7/8C9/85jf5xje+AUBRFFy6dInPfvazfPzjH7/nN/N9/xabLUVRyPMcVVXxff8W2yvHcfB9/04vc4gkSXjttdfu+f1vRxzH39PzfxAQlwobm5NbSsQS0BF1/N3iHX3GNJGY0RNaDUEiNC5td1GQ+Hbq4w33WHTefHOKS4VzexGv7vhIAlq6QkOTkbKEkzUNxlucH28BsOZLdPv9w+fKQCJkXlVzru6NCVJwLQ1HV5hzZE42FZQ84IXzO4j9z+pYNle3erQbLo5agJAI4hhDLlAUiXpdZ6GuU1MlumWJIlUs3vPbPoJqDldIlWORF8fsTBLm6pWVop9WJXgkjaNTDmuDgH98eZu5pkUQZ+iawsYg4H97+ijfvNpHQuJK18fSFLwk598u95ipm6wPAlY7NTquwT++soVjaEzXDJK84LXtCT/70BxZKZAomaqb/PMrm/zkqVmu9QLOzFUs7kGY4ZoqcSro+gkDP2auaXO1G5IX1UFCUxR2RzFCkdgZRUwihQ+fmOafXt5iteMw2zQ5MdXBTzJe265EOeqmuv+fxjCoetQNS2OjH7KnVef2tmNQtzQEcGFrzFNH21zvB5yeq5GVKXlRKakJKta8rsj4cYokuby6PUFBwjGq9oQXp1zay1lqWtRViBPBRm/E6rRLzdB4eWvAU0fbbA4j4jTnv726QV1foianXFzrk3oWulxWMqrrQzb2fGQJTjVVrvWqw0KeKjTUAkMqUE2BKktc3tzmnJsxa7zeG5ZCwZaXkeQCQ5VYqGnsrg1uIUj63EqYfCfoCIUwK7E1GTEaEMfxYdJx++/Ov4mz3yCRWN97475opUPaxvuHdHsfd8dbBuWPfvSjTE9PMxqN+MVf/EWgcgxZXl5+2292u/9kWZaHbiDvxJvSMIwfeUMKIQRWuypnHbja1EyV1bkaLUfn/Pnzb/szDoKU8MaQohSc2xyzMD0FwOKMC4bK7ErrTctz64OQRjakk5pIVCXrrCiZalosLEzjmvphP8yOUs4NpMNDRSkE64OQ6XaTVr3O+e0J/TBnc5Cw2LQZ5gpFKbO4tHTIcj13dZOZ6WnKouQnTi/w3NqQshQsdBweXWqy0fOYqln7c7cyQpJI0orw5MUZfpzjxTmFEBiqzPR+FplkBQ/M1mnXdIQQnN8c8shqhyvdgKZVBWkZiaIsudLz2d2XWixKgaOrrPcj5poWYVaiKwrPvrrDzz68gKlVfdymrXG87uDHOaoic7k7Yanp8NLGEFlSObc5ZrZhcmXPY75hoysZjq6wPorQZIW6pbM+CDFUma6fsrs/f73Utgmygg88sUTfi/nm5T0+fGoKhERalLyyNWaxbbE3iZmq6eQlPLCvilWVhUtAIMuVUIciS3S9mEJU43NhVuIaKtOuyVo/ZLljcanrc3q2xp6f4Mc5ti7zU6cWeGljQpgULDZtXtwYMeUafHC1RZgWOLpCUggWdJ00h0dXmiRxRFC0udaPCJKcMC2QZJ1L/YT/+ewMS3OzmKrEsZkayy2LUZQT6cOD1cPcTEEQ56y0bZ6/MSAXEgd5p2nDzMwsDyw0Du8dtidMdn20rMDQFBZmXU7P1RhF+bvas32ne88gSG/6vK/jgdU3vyfv9P738YOJtwzKjUaDp59+mqeffpp+v0+SVD2vongjE/Gt8IEPfIB//dd/5Wd/9md54YUXOHXq1OHvHnnkEf7kT/6EJElI05QrV67c8vv7uDMOyCXTNYPz2xP2/BRZhufXRm8ojd0rDspqF3e9w2A55eq4RjU/fEBEuRvhxDHUat4VgRcXdP2keh0hOL/tI0kCPy1J9311T806rA0i0rwkzUseXKzjGgqlpjDbMNkLJhztVFaJPT8lSHN0TdmXqFSwVVA1mUFWkZ8+cWYGSYIPH59iytX5ry8XXOiFZHllV3h2zmV3EhNlOQ1TZRIm/Nwj82yPI0DQslU6NZMbPZ9MwHRi4GgyDUdnreuz0ra52gtYH0V4ccbDCw3mmxZpVpBkBZZeCYB4SY6lybi6jB8X6KpKkpcESYokyfT8hLyArhfxEyemODbtkJdUDO4gZb5h0fNjTszWmasZ1CyVlzfGnJmvsz4IWG7ZOKbCYsMiLQqOtC00VSLOM651I3bGIasdl5m6xSDI6E4SRnGGLku0HY3/6ews//DSNmVZUjNlnjzSZteLaVgaUVqwPgj4wEqLPT/l7HwdPy3I8oyljoUQcLXn4xgKTUNjexhDmdE0VfKyGgUKs4LNcYhrashyVTJHgK2rXOsHdByjcsGyDVqWynLLZqcXMeNq7E1i5P1AWDM1JAQ7k4Sen3FsxuXSrn84+rXUsvZH+OLDEb44q3rScZaTFgJNkVlumszVzcN1PgwzNkcxNUM9NI3YGkUEaXELOesHqWf7Xtmt3sd/HO6Z6PVbv/VbfPWrX2VmZubQWel28+q3wic/+Um+/vWv80u/9EsIIfjd3/1d/vzP/5yVlRU+8YlP8JnPfIZPf/rTCCH4/Oc/j2EYb/sD/SjiYMwjysrDzQWqXlNHvH2lroNAb+kKUVqg75NqDjLTm91u7kQ4adkaq22LcZjwresDdFXCVGWOzTic3x5hGipelOPHBf92pc9TR1q0HY2GrXOkbbM1jgCZOM/xk4KiBEWREVQ60buTnAs7HrV9tyg5LfmpB5rcGMQk+xnPmVmXM/N1hkFKVggWGhYgOKPVeO5GnynXZL5pYmkqAz8hTlMalkbHMXBNjZfWh7RrZtVfHscI4JHlOmFcUOyPoLmmhibL5EXJpZ0JZxfqNCyN7UlMkOTYurI/riQTZVXZdhym/OTpOb5xeQ973xP6oycX8JOM89s+SV5yZc/jqaNtDFVCSDIvrQ8RSw06tkHL0Vjr+5ydb7DasdkchpRC0PVSLE3hpY0JP/PQAhDQcR02hz6PLrf5x1d2iNKCpBAkacHza2P+10fn+D9++jRJllOUJRd3PbbGCdf7lexnx9UpRYkqgZ8VbIxCLFVBU1VajsrRjsMwTBhGOboq8+qOx0LTxIsLhmGCpauYmooQUJbVeJssy1zvVpn/7jhCkSUUCZabJt9dG+GNJzSazUpfO0gIM0FWFIyijJ6X0rnpcHhAaLx5rS40qqAb5zktW4WgInrpisxy294XE6kQJPlhdelgFlmikkTtuK/vPT9I+vj3auRyH+9f3HNQfvHFF3n22WeR5Xc+2izL8husq272Av3Upz7Fpz71qXf8+j/KuBuxK8zuNhH85pAkiZW2Xcke3uFU/las71NzNbZGAXMNi61RjCTBhS0PXVMIsxRZlg8z6J6fcmTaRVcU5hqVddvGMMJQFXRVpmGqWNrr665uaRUhi6qnbBuVxeCRqcpZSYiqNz0MM/wko2XppM0SJIndUchczcIyVTaHEVujmNmaySgS3OhPmK4ZGGqKF+eM4wJDlWg4GlleMgpzplyd17YmnJqpk+YFrqny8uaYjxzvECU5Tx1p0wtSVBlkWeLZ13ZRZQVTlfjkg3N8/cIe0w2Tj5ycpmFr5HmJo6v8/UvbpLnA1GUeXWmxPY4Jkkqd68MnphkFKV6c8eTRDld2PTRV4krXYxJXPelSVIS8x1bafONql9m6zXeuD1hu29iagh/nDKOCpqUyXTcI04JBXHBjvUuJRNvSudqPiZKc+YZJz08JkwIJhVEUocoyP358Ci/JmYQZL6yNaNcMhqHE2sDn7HyTG/2QSj0Ljk65hHHOI4sNnl8bsjGKOTHj4hoqe+OEjdGQ6brFwE85OVtjZxyiaSot16JTM/nW9SGyBEGacXTKYbll48UZpiYz9FO0/YPiQdVGliQcXWEQZKwNQsKsGqN76kibhqWi77fJRtHr40b2TY8/4CfM1U2cfUewg0Ctyt8/w4iD6tIgkRgE6TsKqO+V3ep9/MfgnoPy6uoqSZJgWe+sJHof7y7uxiS1tXd+iHqzU/nNh4Cbs429SUzL1hhFOeOk5Ho/PJR71JUSfxTx8HITL86rjXDfJ/nA4i5Ii8NyfHcSo8kwWzPYGEbkQiBT+QMvNC1kWcJUVXa2PaKsZKllsDeJDw8LEoK8KHh1Z8ILG2MUmcpnuWkhqARDTE3F1BTCLGepZaHIEjVLw9RcvnG1T93S2RpFHJ1yqGsyJ6Zd+kHC3jihBMZxRttRcQ2NrXHEqzse13sBqx2bjWHIyZk6iizRtHQu7o746OmKjd2yNTRVJkwLbgwivLRAlKCpCjf6PmdmG7Qdjc1RVda/sDdBlWVGYcrylIOlKuxNErK8RNcqcYu+n7I6ZTOOCmYbkOYCP8mRJTg9V+Pynl8ZdgQJx2dqdMcxDcug68X4acYozEnzqjUwCFIM1WTHj7AMld1JzHSs40fV3zgvoeEYdL2ID52YRpIEj6826U0S6pbFla6HJsn7PXqTmqUy5RjYmswrW2PCTDDY8RACoixnoWWzZOqM/JRYrqRLXUNhe6zQ9WIMReZqP6TrJRyfcUkmJS3H4APLzcP1n5ccBtisKClL2B7HTNeaKLL0Bsc0CQiSSof7YB4fSiax4Eo3OAzUK20b+/ugDX9zdWl9zyfSh3ctjd+fRf7RxT0H5e3tbT72sY8din28k/L1fbx7uFuvSYwG39Pr3u1UfnAIkBC3ZBvFfgbr6AryPqP5QBs4LwVLLZsir4hEIDjStsnyEkNTbnndrpewMarKxooi89BSA00WRJng1a0JV7ohU67OYqsqjzqG+obsvSKqedwYBDRMtbJELEvGUYKiyMRZQV6UzDVMtgYhR2dq/O3z68zWTDRV4sHFalba1lVURUbVJP7x5S1WOi57UoIqyWR5wU8/MMcwionzkvmGxa4WsTeJGUcFa8MBuiLx9NEOWVGpRN3oB1zuhuRFyUzNYKFpIcoSLynRFImFpkVBSdvR6Xop13seZ+ebeFHCT5ye5sW1EcuLDV7dLkhzsf93AEmWkJGR97W5TV3mzFyN7XElvGJpCjvjmOmaia7KhFlBTZYxDQVTU/CilEeXmxiqTNPSGAQJdVOj68XoqkyUliR5ZZ+ZFwVzNR3z6BQXticcn3YZRBmGrlAUgo6jY2gKNwYBZxca7Hkx4zBlpWNzbLrGt68PkCUZRYa5hsXAT5hyDbJCoBSV6pcEbAwCJElmFE7Y9RK8OONKN2ClbTMKcq72A9quQcvWqJnqIQdCUyrCnixVh0VLV/CTgiDJDzNUP63WpaFW31kVlCut8IPXEUCY5rxZOLzXAHqvmgL3Z5F/tHHPQfmP/uiP3s3ruI/vEXfLat9svOJ7wcEh4EY/PAzIB2SwA1UkS1VoO1VGKFNJHs7UdGqmzu4k5oG5GuMgY7FdZUU3l8bXhyF+Uhz2iL043/cUllhuO6wNwn0VMH3ffEFwox9QlJXjk6BSfBpHKbahsj2KEaWgaWu0HANNlgiinKZjs9YPODZtsTaY8OhyE0mCvXGMhOCx5eahsYIXlxTIDLyY4zMukyhjynUZBjFeWqKrEn6S8shykxv9iggmhODUXJ2+H/PQQoNcQCmgKAscQ+e76yNGYcqHT0zx/13pY+sKkzhjvmFyaW+CnxbMN22ksuA/PbpEUeYstm2KIsc1NS7teLQsnb6fsNg02Rz6/MTpGTb6AZ98YAYvylgbxOhqdWA6MuUw8FP8OKMoFIKowDZVGjWFnzw9zdogZH0QMlMzWGzZWJrMYsOg41psDiNWOy5Xuh6qrCBJEmGW07A0hvva12le8MhCk7mGyYsbQ45P2YDETN1g4KfsTarAuti0cA21Crz9gMdXW7i6gjB1TF1Bk8BPMqbqBue3vYr5bapc2gtoGCorbRvTkLm6F3B8ukbb0TkzV2PXS6o1o1a65T0/QVCyOUyo6Rpbw4CNYcRSy8LS5Nc9rdXqULjnpcw2DCxdO1x7rqEQpAXtO9wHbyeA3quC19sVBLqPHy7cc1BWVZU/+IM/YDAY8DM/8zOcPn2axcXFd/Pa7uNt4r3sNR0cAkohGIbp4eZ1QAaTqGZDJareGQKWWxZtW6fp6Ji6SpYVLLZtTs/WbjGC95OMzWFE7ybNYEOVONKxUXStIoVZdYK0YK5m4I0UXlgb4iU5m8NK9KPjanhxDpJg4Kc4usJs3eTVrTGyDJosE+Ql372whypLzDRMBkFRZWx1E0NXkWWJshRc6wc4moylKXRcjY5t8P++soNjqJyZhbQouNINcAyVi0nOo4t1HlmsM12z2BpGlGV1YFjpOHhxhixBWYIfV/1vSVLYGkZ84oE5NodVQHxpc4RjaCDg+JSFbaj8399dR1WrfueTq02mbB19ucHYTzk63aJpaqy0Lf77hT0kqcp8LUOrsnxFZndcHRKW2jbbk5itUUjN1Li04THfmMM2ZGqGWomAIEjTkkGRcnq+zuYgoBskXO4FOJrCfMviWi/gWi9EUWChaePFOU1LJcgLGrbK4ystrvdDhmFKHlUVlRMzLoYC37g2YH0YMeXqPDhfo2Yo5EUlFCMBQVrQ9VMapl71ob0EDJW8KMlKQUml1qVrr/eVW47OA3O1SroVwThMqVs63746YBDmzNUMzi7UaTk6G8OIhUZlcHJzT3mlZZOX5S2MbIE4HNW7PRO+awCtGbDf5jl4zr0qeL1T+c37+OHAPQfl3/zN3+Rzn/scX/jCF3jiiSf49V//db785S+/m9d2Hz/gkCTpkMh1OwQSBXBi1mWlXYn0K4CiSBSlRN1QEPubUc3UbivfwcBPqnLzfq9vEuWoyut9vXGUszYIibKCc1d7PH7CJMkFl7sefliw1DEZhwkNS2cQJMiSzIsbQx5ebDJTN1jvR4yCFNdQcQ2VC9seAoFr6tzoBxiqwpRbsauTLOOxpWn++ZVdapbKsamSKdcg3/d//verPRqWhrrPxD6/6/PE0Q4bg4jZpsGNXsCDi022BhG5KLm057PacSjKkilXx9AkLE3n0p5Plhf0vARBVS5u2hrjKOO17QlJAdN1nVGY8OLmBMfQkJF4YKFBUQjObY3peQkLbYdRkNJyTF7aHBMlBUlZlcanBex6CRd2fOqmSseWmatXLkyv7ngESU7T1vGTnGGUYigyN3ohMFDDqgAAIABJREFUF7tVP3pzGHFqrsa5zTGnZ2sE+0pnQz9hqWnx2s6YmbrJP7+6y8mZGmu9ACQJL8nZGsVsjSJ+6nSHj52eJkiq7y/JcoK0UiIb9QMyU+f8rseZ+TpZXnJjWJWUFxoWjq7QsDTKUjBdqyozh62Um6pFO+OIvKiU5WRFQZIK9vwEsxdwRpMxVBVZkt7gBKXJgrptHI5ECSoew+VdD7F/4Lw5E75TABWIfUW618dGl1rW4fjWW40zvRP5zfv44cE9/5XjOOZDH/oQX/ziFzl27Nj9caX7AO7ey5aFoCxBliSatsYkynhtt5pRFkJivmHScXUcQ33DZiMLQdsxKhEQUWlvf2ClSctS8FNxSOipHIWqcvCLGxMEAj8u6AYp863K83d7EqMpMrIs0XFM0lxUQWYvYBymnJhxMVWFG4OQHz/R4aWNEcr+hMGD8002RiE//cAcL6yPeGSxhmNqTNeNihmugKVJLDYtJEmmaap0fQlFkhgGMbZRaU3P1S1u9H0MTebkdB1NVRAC/Djlgfk6DVutGMWOSpDA9jjHT3LyEvKy5PScy9VewPYkpmVpFEIiSQt0VeH8jse57TFn5+skeUGO4LVtj7qlcr0fsj2OiLMSW1eQUInTgiNTLle6PkemHDYHAafnG7ywPkZXK1ekS12fYZhhKDJPHmnB/t9SV2Rqlooiy8hSVQH5wEqLzWFIAQzChMeX2+yNIzRVoe+ntPZ9llVZom6p1AyNQZBzuRsw8jOOzriVg5eo2PFQVTEMVeXqXoChSkzZJo8tq2R5wdPHO4yClPmGyZF2xcq+OagdVIuCJCcrSnJR9YolBFFasj6M0JTKL7rjaiw0LTZHMbZercGFls2pWfdQPKQU4paADLeWku8UKP2koCxBkd/4nINDg5UOeWC1dcf+8/1Z5B9t3HNQNgyDr33ta5RlyQsvvHDfKOI+gDv3spuWyvNrI85tjhFQ9e68FNuQEcjc6Ifs7PeUz8zVaVrqLWQZLy3w45RHl+oU+xZ/WV5QNzVOztnc6Accm3Fx9UrJKswlyjBhrm5i6wrzTQNVggToeVk1ixxWc8gXdz3OztepGQp+LLEzjnl8tcnORCZJcx5ZahKnBZoqMYpjdicxMzUDSZKYb1m8vDFmc9+LeLHlEKYF3UlClJVsKTKn5xwapsb/uNhjZ5zwY8fbTNUMRoFO3db4ztoQQ5GYbpg8fXSKK7sTwjTn1IxLL4ixNIWNUUyWl7iWxskZl0FQyX5KsoSQKmOOjq3RtnVqhkzdMqibKo8vN7kxCJhE2b4ASM5yy+b8joetq4yjjB871gZR8vBikzjNeXSlSc9P2RlHPHmkw7V+SJAUKBLUTIVhmNJxdNK8pO+nTLkGaVYFPCFAlSXOLtRJ99nO310bcGahwcBPK4/jPCPNStKipGHpqLLEC+tDdFVhyjV4dXtMUhQc7TjsejEG0LR16oaCJFfX3No/fC20KpnShZqFplVleEevLBBvD24HAjbS/vqzdIVdL6Fhq2iqjLtPDDw65TLbsN5A0jpoA60PwlsC8gEOS+Z3CKAzrk6cvdEd7eA5bUenbYi7lqLvzyL/aEN55plnnrmXBz711FP82Z/9GZubm9y4cYPf+I3foF6vv8uX9+bo9XpMT0//hz3//YD34jNKkoS1X1a09jfJqz2fvBSEaeWHvDGKmG1YCFFSs3RsXeHYtFsxfW2dtUHIhR2Pvp9yresR5iXrg4CXNid86/qQPT9FU2U6js5Cy6YfJGyOYvYmMUmWkksK1/she15MmBa4hsZC0+By10eRJXRV5siUzZ6XoMoSRSnouCZenLPStjg566IqMt++NmBrFDHlmri6ymrbYRAmGJrM8zeGhJnAi3M0CZqOgR/n1EyNvChJS0HL0XFNjcu7Pu2axcXdCYoso6rVXPY4yqjbBhd3JmyPI9o1E1NTGUUpx6Zd+l5CLmC141Sldz9hd5zwwSNtJlGCrVeb83Lb5sLOmJMzdYqi4MGFBsMw5eGlFpYmc3K60pxeG8ScnK2z0rHQZQUhBP0gRVcV+kFSZXqSoGbohEmOrkp0XIPVtkOUFsRpwQMLdaZdoxJ1ERWD/vSMS1KW7E1i+kHGfN1ECIFjaNh6pbZ2petxdNrlajdgFOZ0/ZjVtkO4r3q2NYpxTQ1dkTg65TIKMvIiZ2W6wUrHYRhU2uIygvmWjaMpnJyt4xoyXizoTmImSc4rWxP6foJrqFh6RUAztaqVkOQlkygHBEtNmyMdh7mGwZSrI1FpfM81zMO1e3vgK0rB9viN7ZnVjnP4+I5b9bPrlsZqx2HKNdgeJ3d9Drz1fXn7PfX9Dsg/Cnvf+xX3nCl/7Wtf44//+I8Pf/7Sl77EZz/72Xflou7j/QshBDvjiL6fUjdVdFUizgraTrX5Jnk1fmKqCrauIpCq8aebMo0SiavdahRpEuXYWiXecaMf0nY9oJo37noxaSGYcgw2/ZyaXpGg8rzk4t6EhxYXeGKlxcY4xNQ0irLk4YUGYZbTtnUmccYHjjQ5NeOiKRJb45izc3WEJPHK5oi2q9O0dZZbJkcck29eGxJlZZV5mRoXdipjgOs9n1NzNZKiRJUrJ6ITczUu7nrIksz57QmfODvLziSm7RjkecGpuQZ5njPXMNnzErpeQsNQWGg7bIyq7yMvq7GocZyxM4lYbDggC+qmQpwVHJ2ukRUFj660eHljiKZp/P2LmxzpOHxnrc8Hj7S5vOujybA1CFloOXQcle+uR/R8H9fUyAtBw9ZJ07IKpBs+uqpwaraaBReSoGnIhBE8eaSDQLA9iggUidlapYiW5lXvdHMUE6Q5WiAzVTOYcg2CKOH4dOUyVgqNcVR5MctI6IpM09ZwDRVbl5mt62RBzqm5OktNk+mayZU9n0IIdkb/P3tvFmvZfdf5ftY87/HMY82DXbZjZyJ0mpCmu8UVfR9oKYi88IAQAgmEiARIgHgHKS885AV4QELiAV2pH26k5t6+9xI6nYSQxHbscpXLruHMw573XvP0vw9rn+M6NThlB1cc+3ylUlWtc85aa+/9P+v3/w3f7zdiuemgKxJZCTuDkLZr8OpbXUZTO8nDcczzKw0uTy09Ly/WmK2ZdMYxd3oBb+yMGEUZo6jSxm45Dw9fPUhvaljqjywlPzhgKYQ4LT+f4n3jfblElWXJrVu3ToPyxxyPeoDdOvDZ6kfsDkImcUGSF8x6OjVTIysE0nTOda1lT6lLnLDTg6mgiCKTlwJ9SlUxdZlxnHGvE6BKlaZ2pTOtEsWCPIfdUcJ8zUCzZFRFYhRWQzjrLRfHUCq7P0kiz0vu9QNqicZCw+Tl7TFhUnkAX5x3+eHWENuoBEEORlUJ+5c/sYylyehq1SPe7AU0bJ22q6OpCm8f+qy2bDxDRZMFhqqQZCVpITBVmawoWW2aOIbOOMnojWMMVeHtwwkbvUqXe6Vu4Zo5zy7X+MfrByiyzO4w4rPn2vzr3T5t1+RgEvHvz7dpuwaTOEcyVP6v1/dYn/G43QnoTlIKAS+sNnhlc8CFOZeWY9ByVHRF4u1Dn1uHQTUBLqqhJAWJl7cGaIrC+oyDokh0gwRFUgDB3jjh7iDCjzPWWg6TuKBua7yxMwJZZtbTeXV7RJjmlX9yKZAkaNgGG92QnVHMuRkHSYKaoWJkBZMowzNVXFPF0TXuHPqszbh4otocKIrCS2sNZj2Drp9wfsYhSDJKJJI8Z7Zm8MbuhFuHk+kEOzj6mIatM1e3aDn6NIutWg9bw4gZzzye6N/sh1ycnz0RKB9Hb7o4V2W4PT+h7RqsNq13zVxPy8+n+HHwVF2iTvHTg8cJIhwd95OMvp/Q89PjnlvD1hiFlcTkctPmm293KQU4hsJ620GVoWbayDJkRTWwtd4ymauZJya4TU1lxjVQJInDcTIdvBJkhUBXJdIM/tftHofjBNdUaJoSHT9jtWXhGRr3+gFZXtB2NGxdxVRV2m6VuZ1tmghFZnZvzO4w5tXtAVkhsHQVJPDjglnPrKQQg5QoK+kHKQM/4ZmlOt+50yPMCiZxzidWGmiqxGLdoB9krDZNFus6dVOnEWUEkYOsqni6iiIK1po239sc0pmkXJx1KCW4fTDGMQ1kE/7lXo/Pnmmy2LC4NOeiawp5UfLq5hCQKMuSvIDtQYyqSFODi5wwF+yOYmxNYdYzSfKCjW5ImFZ932+8ecBiw6Jp6jyzXOOLlyDOSxxdxtVVvnOny9kZl35YSWs2bY2fPT+LJEq+c6dPSbWRWW05lALGccreSCIrBVGScXHOoTdJQVMJkmojlhYlnqUhEERpzp2Oz+WFGpvDgHnP4j9cmSPJSjqThL1RTIlge5Tw8+dq2LpCWZb8cHvIZr+SW1VkQcM2mPcMSuAHG332RhHH+zkh6PkJwyg7VpU7CoKVxrXEctOibuvH/OPWNGAf4VH0pp1BiJ/kx9PY/SAjmqrO/ajAfCqFeYr3g/fkEnV4eEie5wgh2N3dZX5+/mnc4ymeMh6XMVyad7l14LM9iJgkeZXdTMuAAonbhz4NW0eRJWxd5vKCR1FUU8Lf2+hTlNCwtGlPzyAXVd7ceGBYxjUUXlprMgxS/LiismgqrNQt5lyTnWGIBNRMjTjLuTMJubrcZhTlDKOU3WHE5fka++OYs7Mem8MQRZVwTQ0hV0pP27qCYyg0bQNDSQnTEl2tuM1BkjMMUlZaFjVDZc4zyMqCF5ZrXJ5zkWWZN/fGZEVJ09E5O2Nzfg5mHIv/cfMQVZKYq2m8sNbmu3e67PRDFhsm7RKWalXJ1zJUdgYha22XohREWYljqFi6xpt7Y9queWyDOEpyLsy5vL4zIc1Lwqxk0TbIC4GERGPqc1wgiLOCplO5MrVcna1BhCQrhGlJzYTbXZ+aqXF7a8icp+MaGp5tcK8bkhQFkVEQ5wWrkUOaZby41sTSFV7ZGsIkZn8U03QM7nV8/t2FWV7e6jPrGlzfGdF2TdIiJ4ing3JBQsM2+PQZnTgrycuMT6+1+czZJoam8E9vHnLgJwyiFFNVsTTISzgYRbyxN+a7d3qIKfN9rWUjIXFu1qVpa9zrBqhTBS6JalPVm4qF3NivWhxHgfMd9TnpBP/YNU6Wkx9Fb8pLPtQGFaf46OGJe8p//Md/zCuvvEIURURRxNra2ilP+SOKxwkiWLpyfDyZTpdu9kNqVh1FBl1TSPICW68MAJKsRFNlXt4aoSkyuiLRC1IGUcYvXK1KvTvDmPm6daLcV+kMCw7Glc5xMX009yYxcV6gKDJ1W0OVM2ZrNtudlJqhcm6a7cVZySTKWW/b3O0G2JpSbRakyuvZT1J+sDliqx/QCVKuLniko5BJWFAzVZ5brvPyRp8Zz6SYlmLvdCK64YimrRMmGS+tt3hls083SNgbJfzXF5f59u0uSVbSaljYusZ/e3mLF9daWEZFBbqxN6Ht6nT9jDDN6UxSTF3lYJwQ5ZXJvanL7IwixknBxTkXqSEhIShzgYxg3tORpYoy9srmAEEVXHpBzGLNRZSC1abNZj9AV1U6QYwsSyBBLgTfervHF6/Ms9owaXkGqgzPLtX4wb0+Nbvi93qGyr3OhKtLNQ7GEU1L54WVJi9v9Vhq2mR5wf/+iSV645j/dHUBRMkvPb/EziDk5oFPCcy4BrYuE+eCz55tE6c5hq5yfsbhymKdm/tjciHR9RM0WaakpOnajMOI13fGvN3xmcQ5bVdHIB2vsyDJj13IXlqv88bumDwX9MKU9bZN3dKOVeWOAueTUoweRW9K8sol7UGcCnmc4oPCEwflmzdv8vWvf50/+7M/4/d///f5vd/7vQ/yvk7xE8TjFIV6/jsTpUda1QKOA7FrKMxOg44qV9nN4ThGU6psxzFUsiKdcnRzvGmmcj9VpGlrj+3r/dObGTf3x2z1QwohmHH0SkHKV1ltu/T9hLajc2bGph+kKAogqinrjV6Ao6uoikReVp7GuipVWfMg5MyMS+zm6KqCVBY8u9Lg1sGEO50Az1CpWRqDIMVPcq4t1vm/39jj02dm0DWZuhmyMwixdY20hLudAGPRoxcU+GnB24c+62238khuWIRpiCyrrLas6QAcjMOcq0t1XtsesNZ2GIYZr+2MWG6aXJqvcetgzMU5F1WWODtjs1zX+U6eM45L8iLmv7ywSBwXXJhz0OSKb/z6brUZqlkKDUsnSnI0VWazG9B0Db59u8dSw2Shbk9tLjNmHI3FmsXrewPcgc4wyumHBb1xzM+cnWWrH9Bq2tzYHfPCSoOyLJFlFVWS+NkLcyiKigwEaXpMJPrB1oAsg9WWzbkZB1lUm6zVpsWdTkBWlDQsDVuTEEgkRYkiS0ziDNtQsTTleJ05usKb+xN2BiGztsnleUGal5wtBYt1i/maAUgn1tWT9ngfFbzXWhad+5TljnAq5HGKDwpPvLIajUaldRuGtFqPUoE9xUcFj3vgtN2qdwpViXnG1en56bFu8OoDwgu2rnAwihjFGepUwGEYZkgSuKb6yOu9W5ZeIiqZRk1mfxiRFAVzdY9n2jNcWW/znbsDDE2mYekcTBIkCeY9k46f4JkqkyTndsdnHGc0pt7Jg3GIbWjc6wVYmkLTkpnxTG4fTirFK8fAUCU2+iENS2cYpqSloB8WlfymruCYCnFRIsuCKM1YqJuYqkzTVkAI6pbBVj/gmeU6RV5UVpNtm5eW6/SDFLllsz2MGMeVU1PbsaqMz1Q507KRkVhpWhQluIaMqSn8cHPI/3ZtiTudgCAt+MaNA64sN/iXjT5nWjauqfLvL8zQsHXe2K1K7Z6pHbcFbncqypoqK5gKpKWMYyrsjmJud3y+eGWeza4PskyQlDS9SszlwqyDEIK5msHb3QmrTYehnyIrBm1H49yMTZQVRIlGXlbvkYlGUGQcjGO+e6dPzdJJ8gJJwBcvz7I9jGgYKt0wY2sY0GqkdCYxjq5QFiWSJlMNB1buXtX6kGg4Gq6lMZhy0HVFOsEpvn9dPUmP93Gc+6OWzRFOJ6lP8UHiiYPytWvX+Ju/+Rvm5ub4yle+Qhw/zN07xUcDjyv3rTatY3/lo8GZa8t15j0Dx9QeEl44OtdnR0nVkwTmXIP5hoE3Naq//wF3RKfq+slDWto9P0Gi8sxNs4I4r4aYRlGGaRSstGxeyEV1b4bAs3RqpkKYFvy/Nw5IspKbBxMcXaUoSspS8L2NPlcXPN7YHfPcaoNBkDJXM/GjDE1VOOxXvOf2dOhMVarKQFGU6ArYWpVxHU4iVCReXGux2d/hXjegM474peeXeGN7RC+oTBEWPA1DtVCUkBeXm1zfG7I3TgiTAs9SoRR86kyLWwcTNvohNVNjsW6y0w+5uFBjFOXEWcHd7ohZz+TW/ph4OuS10nYJ4oSGrXFp0SNMC17fGaFIEmszDq9tD1moW3QmMcsNmyCrSvW2IbPZjxhEGQs1g3TqdPX2wYSmrTFbs7jX81ElCT/JCbKSSZTQdDTKsvrMamaVzZYCfuZ8m61BTJLmBElGx0/ZGYbIknxsp3gwjliq29w6CNAQnJ9x0FWJSZpjWgpxWiLLEvd6IedmHNqqzmfPt3l+pcHOfQOBYurd3HZ1HF19SNby/QTORwXv00nqUzxN/Mig/NWvfvV46rbT6SBJEvfu3eP5559/Gvd3ip8A3q3c914fULIs88Urs1yYd+n5CS1HxzNUwulgU8OqlJWCOONgktDzE3YG0bHr1HLTQqKitvSClBsHE/7l3qBSk1Lg6qLHjf2Aqwc+Vx5xb/e6Af/zrS6TOCMrBMMwxdRUDFUmyUqCtGC2ZjKYJDQcg4ZVbUj+28vbKHLlvbw7imjaeqUYpqncPpzwn59dpDOO0BWZvATbVPlftw5Za9pcmlcwVJmdoc8L6w3Oz7v0w4xXNodYhsZnz7bp+yHnZlwEMm8fVv3T1ZaFocK5GYfzczXmPZ04TVlqWvSDmCSD63sj1lsOG90Jn78wiyRLWIZMWYCuqlxdstkdBHQmGW8e+Fyed7m5N8I0VHpBiqlWfetnlmqoksT2MEJRZSTAUBX8OGd+6n/8zGINP8pYbTrsjyIMRSZOC3p+SmeScHnBZbFhMYoqvewwK7jiGZyb9fCTjJu7I/7x+gFhWgLl1CIxBSQWGyZfuDzDD7dG7I5ipCnNbcUzudevBvmeX26w2raY90zmXIOdYUwpKlrdUUYsEPhJwWrDqtaKVPXY/y0D5+kk9SmeJn5kUD537txDxy5duvSB3MwpPjx43IPo/TygZFlmve2w3naOj7U5OeVdeR+PmPV0VlsWW/3o2JrxyryLrckURclGN8DWZLKy5NyMyzdudbCVgm++1UGebhruv7ckK6ibMlkhk+Y5siRhaxLLdRPHULgw51Wc6iSj66cYmkxeCBYbNvubA1S56n2em3G5tlRjZxwRpQVhkmAbKnVboxlrjKOcbpAyqyi8vjemPX2PbC2lF6YowGrbpSwFaVEwTkpe3+3z/GodWQIE2CrMOhYHkzFRmnH7YMyMp7PatHluucGb+2N+4co8+6MYhMXmIMLSJZZqFnc7PnM1h0mUHvsTX17wWKiZbA0idFkiKgSaIjOMMlYMlb1Jwp1OwELNpG5rBFN1syDLibIMS1fo+ylpURKmOVeXamz1gmkrAhqWjq1ValMH43hq+WjyM+cqA5IkL1mfcRhtDSmm5iKrrRqqXOlDz7gGjq4w6+noioyuyEx8n5ZtoSgyhibjGRp7o5gkL6diM5VBhETlFLU7iLB1lZ1hhJiW+Vca1hP5G5/iFB9G/Mig/Mu//MtP4z5O8THE/f3jJK+muQ8nKRfmHK4tV849F+ddpFLwT292udOZoMoSqizz7HKN69tjklxgygLX1B5JVSmFIMwqqcSVhs3dXgCmSjdKyAuJ/+P7W1iaSnu6GTAVmbmayShIuDDrkJeCcZTz7TtdkARpAS1b5cAv2OhOaDoRyw0Tz1SpWTp3uz5rLYe2reGYMkmeczCKkeUqg1tpmlycc/nB5oiLCzX8OEEgMFSJhZbD16/vMooKFFmaCmek1GyNNw8mfOdOj/NzNe50A860ba7vjfnC5Tm+cavDesvh23e66IpCkGT84rUFbu37tBydvVGMrijsjyrnrc1+wJV5DytM+HcXZuhOYhRZ4nYnqAajKKnZBq9uDbANjU+tNdnoh+xPBTjirMRQKzevYFrSHk89kr93t4ety7RcA1OVOde2sVSFOC/IioIZR2d/nGBqIW/u+0RZgSpL+CJHlmQMBSRVnmpjV59jz09ZrFtARWsqEVya9yrDiJJj/2x4mFcMj/c3PsUpPoyQf9I3cIqfHIQQ9IOUrX5IP0gR4kEJ/Q8W9095H5kHAMRZOeU6q9i6ysYgZrMfIskyYZoTZQWvbY+wDBVJgvWWw6yjMUkyXt3ss9ELKMvy+NzX98Zs9CJu7I9xDW1q0adw56DiA2uqVLlZmSqfPNPENSUuLtQYJwXf2xhwuzPhhdUmaV45MCWFYBhkXJp3WaybOLpKkhV87nyb5YaFpgCSYK5mMQxSlhomfpyzO4zY7Ifsj2KCJGejFyDLGjJQlPDG3hgJ+bj/WhlL6CiSgqmqmJrCvW7A2baDIkmsNh0MRWahbnGnG1AKiPMCS1PZ6AfUbZW7XZ/lpoWgxDMrUY5zsx5b/YCNfswrWwP6YU7fT3lhucliw2bWM7nb8TkYZwzCrCpdqwpBWrDZC7l1OMbQVFSlRJIk3tgfE2cFtw99kqLyHRalYBTldCcZkyQjygrOznh4lsqZtoOfloRZwWY/RFPlaVAVtK2qvH9+zmW5aZHm5bH62xEkqs/LMzUUWZoS5gRFKQjSnOvbQyZxRtdPmCQ5W4OQwX1B+hSn+DDjdK7/Y4rHCYT8W2UUj1MEux/3T8ceUag2++Ex3WqlaSHzThatSDDrVhmTY6oYisLVpRqNcsztXsj1nTFrLYvvbQy5tODx+fNtorzgTNthEKQkScVTduY9giRjsWkzjHJsQ2cS5Zi6yp2Oz2Y/ZhIlXJ51cTQFXZXpTmJUVcY1FVxdYbFu0plUA2maIlOzdQxF8MxSHUWSKIqCgR/TdA06k0qVzFZlun7KW52AhZpBkKjc7QXUTBUJiJICWZYoRUmYlAgqday2o6MqEs8ueuwNE5q2DpJg1jNI0owgzQjSAsdUcHQVW5OZxCWJIshLiTuHE55drvP8is6N/Qm7/ZArC/Pc7UV0JiltW6LhauyMQyxNoTtJQZIphCDJCkxD5cbumM+da1EKGMU5h+OYz51d4Ju3e0RJyU4as9Kw2OyHXFuuE6UFQVpgmwqGYVdDXqLk0mwNPy05HEek09eW5UcbqEod7YtXZhlGOX6S4cc517dHjJPixODficlqBP0gY7MfIkuCnWGMZ2i4poJAYsbVubyQnfaET/FTgdOg/DHF46hHc54BU/P299uPe9KAf/+Ut6AynL8wN4uhVS5Ds16lonSURQskXFPBMlzmXJ1LCzXCNGOvk/PG7phZRydOSg78hI1+SCkEi56JLARnZxz2RjFpXpKXBXOex3+/vodAxjUUnl+pU5Yl+6OEoiiZr1nsjSsZ0U5QmSg0LY15zyArSoI053CSVJaA45irizXOz3ksJYLX9kckhSDKBI6pMAgyapaGKkOUFhyMEhZrJpMkZ7Mf8omVBq6hEmcFti6jKSZ+4lMzFa4u1jiYRMRpiR8lrM/U2OhNiHOBIsHlxRo/f3GWf367SyHA1hSW6ibXd8estWzGUUrbNXljb8KzizWGYc75+RpZnnNuxqEzSVArrRbOzjjIEnTGMQ1LxdYsWq7BnQMfVZG52ws523JoWgqO4RJmguWmTRBPSEtBXlbmIHlRUAJLTYvafbKWiJLdYUQ/rHyK90cJdUvlyqJXWT8KaEghwyinYakcjmP2hhFIEhsdn9magWsq1E2NsiwACUuTOfQTNvsBYkqTut0NiLMSy7DYVuopAAAgAElEQVSRJehPldG2+uFpj/kUH3qcBuWPKR4lECIQ3NyfPEQtea/Z82MD/gP93oemuXWFw3HM9pT2sj2o+rXrTZNxVGVCAol5T+czZ1tcmnd5eWvEDycpeSE4mMSUU3OCsoQoydjJStbaDvQDyqJkZxzz6TMt3tgb05xaL1q6QlmKij+bFSRFZTV5fXeMqclcmHWRpSrz/8Rqg6IoeWPfp+lW/sCaotAPU4ZhwYyjsdqwubE/wtZllusmr2wOKztIR8cxVdI8x1AlLFVh3jPxp9rQqiIz7xpcWHD5zNkWC56BH6foko5SV4kSg7c6PoosI8sFaQ7/erfP58+3eG65Tm/agiiFYLFh0g2SYzUvhKiG0mydW/tjxqHFjKfyX19aIUxyCqpg3HY0PnWmRT/IqJkKwyjlM+tNru+OqzJwkPKZsy2+u9HDVBV6fsz5eYdRmLLUsDCUSiRkYapnfr+s5STOqJs6gzBHkSp6nG1Uxh2jKMPWVe4Nx/RFB8+oPhNFlmg7KgiLt7s+UVpRuX6wMWJtKr6CAF1VWGlapHlRTaf3Q7KixFJlWo7B9e0RzalU5mmP+RQfZpwG5Y8pHiUQ4icFZQmK/M7D6v3o/D5OEexR0oRH09xNW2OzH/Lq9gj9Po7yzjDmpbUG/6Fu0ZkklEKwUDNpOjqDMGMYZjQtGXlcMI4y/LggKUrKsuRNz2BvFOEaKqoi89KZFv/Z0tkY+MzVDVxTQ5Urw3sZ0FUJTVG5uzsGwDVUDicxrqlW3yPl/PfX97m21GAcpmSFwPMMVFnG0GVsQ+HmQcAgTBgEBSUlaw2LT603ebsTYGpVTxeh0A8ygjyn7ejc6/k4ho6tSczUKzOMtw99LF3hTMvm7IzL3X5AGGXsDGOuzHsoisz39wa0XJ1uWDCMEpq2QVmW6IqKa5TV9LqlEqc583WTjV6IH1eqYq6ucGWxwbdud9kfJeiKxM9fnidLM/6/Wx3Oz7jsjWPyUnBzf5crC3UkWabpGLx16PPiaouaIZHkOrcOJlxdrBOnOc+fa/Pcch1Zlh/ius95BkmW41rVEJ+hSlMrTBUhgafLbOxB59BnEmfMeQZznkndUjn0E3aHEa6h0Q1SxlFOKUo8q4mtq6R5CZKErqmossTVBY+lhoWpKby5N6buuD/Wmj7FKZ4WToPyxxSPEgiZc3Xiqab1/XivOr9HAf+IQ3pUvnT0hzWE4Z1y962DCVvT+7mfoxykBast+4QpwNF9AchlyuX5Gq9sDdgZ+rimwktrLb7x1iGOrnFtqepjvrYz5peeW+AZq84r2yNGYcT+UW/TNcgpSbKS13dGrDQsZFlwbsahH2a0HYN7vQDX0Lm+N+TaUoPv3O0yiXMsTaFha3TGEX6SIkkSO8MACYl7roYmyzSnRh3jMOXSgoelK2iyzJ2uz2LDZqsfoSgar24NmHVNGlPhi3u9kG6QVi5WooRSMIhyNBlmawYtW+N2Z0JnkpGXEz5zpsmtwwHXlurYmsJq08KzVPK85ObehJqlsTOImF+u88pmn7WWg6HKgMxrO0M+ud7kTNvh4oJLZ5zyg80+WSkwNZlL8x5zrkHD1rg479ILMibJhGvLDS7NezQsHVWGUZSDVAm9XJx3j7nDQghe3hyiyGDpCjuDynf73KzD7U5AzZraIxoeqiKT5iWb/ZBzsw5pXhAmJXGWkpclflytyTTPsXWVtZZNmhW03CkVTVepmSp+mmPqCklarZWjzd6pdvUpPqw4DcofUzxKCOToofkg3qvOb9PWWG6YfPdun46fHrv8HI6qgBukxYne3lG5+53eMcccZc9QH3v9o+OlgHnX4KW1JrpacVvHYeUwpSsKQuLYP3gS5zQsDUWqrpTlJbahVL3JcUzN1FltWsRZyVrLIUgKtKSgKKs+82Y/IM5VrszDlYUakgRpXtkTTuLqwT+OEv7jM3PsDmIUWeGtwzHPLjWr11YKfrA54GfOtbENFUNTGcWVtrKlVrzgQZgR5wWyVGXrJdV0dpamfP7SLN++3aVm6iBKZlyTlzf66LpKGBbc6QQYmowQgrarM18z2R+G6JrCJ9abvLE7wTNVmpZOmBQESY6lavTC6h4KUZXxTVUhSHPOzbp0/YQwLegFGVcXaliajCxJJFlBAShTpypFruhKj2qBrDbt439vDyL8pGAYpJxp2ZXLkybT91PysqrS2JrMjGdwME5QZFAVmVyUuKoKuQwShGmOqijvzCPMe8iSxCfXmgggSHP2BhG3D3z8pLqfo83eqXb1KT6sOF2ZH2M8KAQihHgiN50nOe9czWShbtF09CkFSfDa7gT1MDjuMR719o4y3vsnsAWQZgUrC94JGc77J7rrZpWh3swkDASTOMdPcnYGES1HZ9YzKEpBXpTM1QyKQnB+zmV7ECBJEqttB4RE29P51u0uqqIgEfHJ9RbDMKVmaSx4JstNi74f0/FjVls2uiJPvYKlSpEsyknzgiAr2BvFTOKCnWGCpso4uswnz7SnWt0leQ7LDQc/zpEQXFl0+e7dPgIoC1hqWNw6mOAaGmleoikybUfHM1XGouTG7ohrS3U8U0FQ49u3D/EsHT/JWG7YdP2YT51p0fEzFAUMWUJSVF7bHtL2TMI0Z6lhU7MU9icVtahV0ymBUpScbTt8ar1Jb5IyjHJ0ReLZxRppIbgwVw2RJUXJvV7AIMwZBClrLYuNXkDNrCwkS0dMPbArbA8iZr3Ku9g2quy554ekmcnWoKJylUKgTdsm1SbOoemouKbGpTkXR1dIsoJ7/QhDhfW2zWLNJC8KNEVmuWlX9o739YmlQMJPJqy2nOM11fNTri3XT7WrT/GhxWlQPsUx3o+M5uMQpsUx1xggKyubx+WmdRyUj3p775S7q4ynNu05vrDaOH7QPjjR/Y6yU0ndrrSXa6bKgmdyMEo4HCfMujq6WpkZKJLEz16cYdbWuHtYlYUBRknGOMmQkDCUSimsM4p4dqXBStPmxdUG28OQ/+eNAy7MuYyCDFNTeGVzSIHg4qzHVj+ibmvcOhhxbbnB3W6XllMJf5SWxnfv9Pn8xVnCLMfRFG7sTzA0Ez/JmHEMLsy57I1iDFmm4epM4owkL0nzqg/bsDX2RxGKJB1POWdFSddPsHWd5aZJmovKRcmwK+tMRabt6kR5yXY/YKFuUZTV99zu+niGzFLd4nZnwoxrcGnO4dJijc+fb9N0dIZhxmrboevH3D4M0KRKY9o2VUSSM18zWajBubbNq1tDNnpRFSjrJgfjmMVGRV3zk4IsK3h5ajMpTUUysyzn5c0Bd7ohQlQ+25fnHX724iy24x0LglxdcLi84DFbMylKuLTgUZRg6zJRVnJxvsZczXzkOg2S/KE1ZagK89MNwvvFk9D9TnGK94unFpTjOOYP/uAP6PV6OI7Dn//5nz/kNvXbv/3bDAYDNE3DMAz++q//+mnd3imm+LfS+X2wPHjENTYe8KYNkrySRryPGqXIcGneO5H5PDjR7ScFdw59ri3XybMMRZYJs4KlhoUQEpIEs47GQsNCk2VeWGtwZbHG1iBiaxBOM/KAhqXy9qHP+VmXtw4mSJJEXAhats7VBY/1GQdFlvjc+RZLDYvreyPuHAYICVYbNlvDkIatUYrK43gcpcx6Fo6h0PcTZjwdTVWQEPT8mPqMx6fPtpGFQG6YIEquLdaJ05JRnHOv43NxzkORAGSiNKXvp1xaqDMIYgoBb+yOeGaphmsoLK03yPKcnUGCaSj83PocnVHITN0ljDMGUY5r6Di6RFIUfGK1yd4o4nCSMjzw+S+fWCJKCi7O1/iPV2dR1epza7kGnzuvs9kPidLyxPBdkpWVP7UsUYiM+YZFyy05P+dgqTK3DgN0PyUrSgZ+RRu7dThhrmZwdiocsjcM2Z9UCmMAwyjj0E/51KLK82dbJ1yaBmFGkObM1Qx6fnqs3nVlwX3XKer7N3uKDI6ukJcwSXL6QfpQMH2SYPte+f2POidAXCqnFK1TPBJPLSj//d//PZcuXeJ3f/d3+frXv87XvvY1/vRP//TE92xsbPD1r3/9dIF+BPDgIJmhVl7LrnEyKDuG+kQZup9kTJL8eGgsSfNjj92j80vILNYNPEulM0l5uxuALB9nRtUfaDg6gyDlyoKHTDWYJETJp8+20JRKIOTFtQaXF2vTkquKQKFmqeQFtJwq0C41LBZKk5qlEmUlr+4MaTsGeSlAgmeX6tzrByiyjKnJvLjaYqVpUgiJOCsrru4w4uK8zVLT4vWdEUFSBaDruxNkSeLT6w0alsGBH4MQmJrCjGdyY3+CKstcnne4slhHiGry+NWtAStNm0VXp/QMzGGEoVSVi9vdgO1BzFzNYLXl4FoaWS7IS7i+O+bqoodn6Sc+g7WWfewMdoTzcy6jMEUAcVoQpgV5UaIrMqOoMqu4fTghzArW2w5RWrA/iiqZz7LKioWoKhOarZCXJaosY+oKQnC8KXxUdaRp6w+5kj3JGjwSGAnTHFmG3WF8Ipg+abB9UrofPCaAN0wE8OrOmBVp8tjrnOLji6cWlL///e/zG7/xGwD83M/9HF/72tdOfL3b7TIej/mt3/otxuMxv/mbv8kXv/jFp3V7p3gEfpwy3YOB1tYVDsfWCeu9+/vVj+pv94P0+Gd7k4Q7h/7xZLhrKMiIYy/no360qkhkheBuN6Rp65UoBbA9jJivW7iGxnLToj4VtZCByws1gjjHMio1rLWWg2dqxw/rw3HM/iiiHyT0w5SWrdOwdUxdRkbiC5dnKYFryzX2hxGyIhGlBbf2x9i6xlLdoG7qdEYxzyzVSYsS15Do+QmrLQdZlsnzSgM6ygS9IOfqYg1zOhX9/a0Bm/2Itq2y1nZIi5IoU5lzTFqOyT9e36dlG/TCqhTdHccMGxavbA3RVZnpTBt1SyPOCiQkvvlWh5WWzTDKKkqXInF9b4yuVFl9XoJnqlyed5n1dEohjgVdmrbGrQOfrUHIMMrY6leVh3vdSg0tSnMsXWWSFHzv3oBPn2kQZwJTqzTIa5aGLMFi3aLrp+jIU96yiSSJ4yz2wQAoITEMM87Nuk9Uybl/DR6OY8bJhCXHOlYFuz+YPmmwfS90v0edc6MfET7AcDilaJ3ifnwgQfkf/uEf+Nu//dsTx9rtNp7nAeA4DpPJ5MTXsyzj13/91/m1X/s1RqMRX/7yl3n++edpt9uPvU6SJNy4ceN932ccxz/Wz/804Md5jTuBYHs6ICMBKy2bZef97eb96d9toRBmJbYmI4Z9bj487P3QtVVNo+dX4hKb/QAhoC/Bi+ttup0D0gJ2Dw5Y9jRsTee1fsSskSPSBD+FW+MhmW9hp0NahkAKBaP7zu2UGWstjSRPK+7s0Ke7H9Hdh1EGb+1XmwE/gqYhc+9wwFLdJI4ULs1Y4PeQgTQUbHQmdPyYxYbDUt3EjxIG45DeyGepbnB3u2CxbuGqgnVL5db+kM1BdR/9UcRBP8Y2DfIs5TAoGQQJs55J26rMLYZRSpTk+HHGatNkdxhiKxJZniGVJaMgZqXm8T/e2KFmKGRp9Xn1w5zFls2dw5S3D8ZYusKFGZOxP0FRVBbrFnuHfXQR4xfV+6xIEi/XbeIsp2U+vAZqhYYlYq60NSaRTzeWeXWzzzOLdXZ7PkEm6IxjdroKliyI4hzFFAwGOUtNC70UpEf9fFVBzQPe3om5dzhmpWVjqzJbh/5Da8NKB7SMk8S9uHxnXZly8dDxVMgMO2MeXG5H5+on0hNdKy4VtnfGJ4KqBLRFDf+gOPGzjzpnIalEaYEhMra2t971NZ3i44kPJCh/6Utf4ktf+tKJY7/zO79DEAQABEFArVY78fWZmRl+9Vd/FVVVabfbXL16lbt3775rUDYMg6tXr77v+7xx48aP9fM/DXi/r7EfpOxsDFixT77/82vND3xH/+C1wzTnIPU5M+dwbnmBJC8w1Uqso+Nn9Hf3mWvMsDjv4pgqI2nCuNBPPDjrTYurFxdoOTpX7qsA2FMVsd1hRF5W5fDVpoWlymyPEiIydpOMrCwZBgmZgE+fW6DhqNiayqfONLA0jY1+xA87XYJCocAgEzKDuODsbI2Bn7PStnA0hcX5BooscXW9SdPWaMxMjsurkhlTKmNajkYhqtctKRpzdRNV09nsjplxKwMMQ1MRSBiawljKyZGJ8hy5lBCKSs228EytKgMD+3f7tFwTU9Po+QlZUdK0DcYJvLnvM4gE857BlcVZesOYuXbVF9/sR9Qsk8aMi2tW2a/U8Jj1DHp7Y7ppgUCQqwpzns7qpMS2TNxCxhYwSQWaYSJRcnm5zTjOaDddMhlKVeETZ+rEeYGtKbQcjY3tHZZXVpgkBTMti4bwTmheA1xdb56oqLy5P6E3iECHCFhpWFyad7l14B8fL0qB7GgsNa1HnqsfpET64KG1eP+1jq5ntR5f5r6/umQLQaxPjnvgUN1HmBUMO3usrqw+9jofND7qychPM55a+fqll17iG9/4Bs8//zz//M//zCc/+ckTX//Wt77F3/3d3/FXf/VXBEHAW2+99Ugv51M8HbyXMt0Hfe0j/nKcVWVfW1cpSkHHz/AMFVcT1KYBZNbTibKc1ZbF7iBCVWWyvGS1ZdOwquX+YKm8aWsEaeVypGsK93oheSFoORqqotANEgAKAXXb4NWdEVcXa1hawd4ooShj8hI0RSIvBYYmM4nzaT+2xLXUSsij5Uynit95Hy8vVAGuM46xdQnPUojTEs9UudcN6AUZmqpwfsbGUasHvmOqlKSst2zu9QLWZhxu7o2RJQlZEpiKwvYgZL5mUTM1DFVitWVTN3RUuSDKStYaJkJI3NgbIcsSYVZwMElRFJ+moxMmObc7EbujeOpQJVMzVHpBCkJwYw9GUYIsCQohYWoKPT/l6mKNJCvoIaEpgi9enmMSZzQdp/Jxbto0HZ3Xp9nmTE2hpCrrelb1eewMKi9tVRLkuWA3TI+D6YMUvceVnS1dOXFclcHWVfykOEHJOzrXo8R0HkUHfLf5h8czBMRxYF5vWQjgXzq863VO8fHFUwvKX/7yl/mjP/ojvvzlL6NpGl/96lcB+Iu/+At+8Rd/kS984Qt885vf5Fd+5VeQZZmvfOUrD01nn+LfBu/VwelJjv9b4sFrHPeL7/Pv80yVIyqsLHHsErTatJAkiSwvWGyY3O4EzHg6QZJx68B/5EDNMKr8d48Uw7p+ws4gombVyYuCtZbN3ihCV1Q2+yELNRNVhjAp+N7dAS3PQAKKQuCZKpM4Z75uoEoSnznbYhxnmJp2wvf3/td4OI753r0+++OY13fHuLrKcs1kpW3x3JKHnxRoqswwymg4FpIEq02LXpDyybUW377T5fmVBlFeYKmVvePnzs9wOInp+hWP+IWVOmdaFv2oYLGmE6c5uaiG0cpSkBUFgygHqXJdurnvUzNU8qJknCXc3Buz4JkAHIxCemHO/ijk7KxHGGeoikzDUpnzTAxNYr1p049SPFNl1msCEmGaARKHk4QwK7A0+XhwTyAYhCnDTMYfRNiajK6q2DrULI3lhlXx3qfB62jeIEjyE0HvCD0/OfH/I2rUctPGmQrS3L/u3wsd8HEMhUf1wI+8n2VJOjl9Pagxs+CdTl+f4iE8taBsWRZ/+Zd/+dDxP/zDPzz+95/8yZ88rdv52OL9ODjd/31PY0f/4LUFEs8tV3zUIzUwhOAHU/UxWdWOxSE0TaFmKKSFysEo4upS/bj8+biBmocyc005nuw2VBVNlrgw56IrMqWoeMJCVBm8rHBsPegYKjYCS1dZaVg8u1znxdU6bx0G7AxC8hLSPGPOM/GnKl4Iwc0Dn1Gcs9GvdLrTXFAisDSVZ5fqyDK8fRDwuijZG0ccjBPOtFz8JKHl6CiyjKFKLNRs7nYC5momDVNltdWkKOCltQamJvOv94b0g5RSlGz2I3aH0XRwyqHpaKiKNPWH1mi7EVmeUZ/Kct7bGHB10cPSFFRFxtYkokTwxs6IT6w1COKcW4c+W4MITZE507Z5bqXOWsvGNXWEKPk/X92j46eYmszBOKZmqJyfdXGnA3M9P2FvlBBSDc0pcpVrKjLvlOIfWL+TJCfPS1qOdiIwt12DfnDSQ1kgvetA1Y9LB3xUdenI+3m1ZZ84bsrFQ8dOcQo4FQ/52OF9Ozg9xR39u137qHZyv/pYkouqFDqlXAkksqLA1N9xKDrCo8rvD2bmrlHRt6rpZ8GsZyIQzHnVRLCpKXiGUpVy4T45SInzcy4zrsmVBY+moyNJEpfmXfwk5/bBhEFUZexrLYeWo2HpKkmak+UlQgBI2LqEosq8vjemmL7Ww0lCmufULIfEFrx9OOa51QZxVj3c3+r4PLMgM0wyRFTRqmRJZs41+NS6xHzNYqGe0HBUXtseMwxT4qyg7erc6wUYqsti3SbLS0Z5yiCsJqA9uaDh6LimxlLdYn8cc313xM9dmqXtVoFQk2BrEHEwjmnYOrqqkBcBy80qIDdtjc1egCJXJd48LzjTsgmn0pd+WrLWsqhZGo4UEwiTKCmYJCWeIR9/RkKIyrRka3jMm3YNhd2p+poyrZxUsp7WQ1SuD3pT+ZOsLp3io4PT1fIxw/txcPpJUDV+1LXvD9xqOmIgnBMDQYaqUDximPVRD8gHM3MJiU+fbVbc3sMAe8qtdg2Nn78yyzDM8JNKC3qtZR/LQaZZwQurzYfkHodhxtYgIspLOn6Kpcls9kNqVp1JXPGttSl1SYhK73p7ELFQtzA0hcNxxO2Oz6WFOjuTAk9XkTyDq4seCMHeIObcjMv8tJLQmSTIUmXZaBsKh5MEXVOQZehOUg7GCUFaUArQVZmVhs1q26EsSkoh0QsSJEni1e0hL6w26YwnrLUcDv2Ycvoevrw5JMyqQTlZcVFlCcdQKKudBaM4ZxBm+EnG4Tjmta0h4zjDMRSaps6sp1JS6VvbusrOoLLllPKEuUaDzX5IklX935WmRcNSH2tastS0WG5Ug233b+Ce9qbyJ1ldOsVHBx/LoHzUU+0n0iOVfT7K+Cjt5o8C97xRsNDwTjwMjwZqHseLfvA8jzPnuN+ZKswELy66SJKEn2ScbVvHClOeIbOyUKmQwTs9T0dXuLk/5vahT5IX7I9iaqZK29UrWUxdYbVlkxcl61Pdb0evrr/SMHENhcjUKEvIs4y1lsfeKMbUZJKkIC0rMZQgLXEMDV2WeXGtSV4ItOmQW5yXyJKEn1RDXppSCanYuoKhKti6gqkqTPKS13ZG1C2Vuqky71WSqM8s1gmSHFNVWGna/MudHs8s1UkGOTISr24N6PoZe6OIyws1ZKkaZlOoNhm7g4C0FGz0Q3RZ4kCKKKgx5xmstWyCtKj4yppCKTiWxVxtvdNH/lGmJQt1611tQY9+/oMMzj/J6tIpPjr46XsS/5i4vye1degT6YOPlaLOR3U3/6iHIcB83XqiB+SDmflWv9LGPhLSONJNDpOc1bZDy9FZbdqPlFB8c2/MRj8iyQsKIShzgYxAUyTSvORgHKMpEoYqV/rOizWeWayxP4oqsRMBG70Q11SRkJh1dc7NOKRJwEYvBElwdsalRDAIMpabFitNh7qlEiQ2/lTzuchKJKZqap5B10+4uTuiaeuEaUGWl0gStGwdRYLDcUJnkpBkJeszNiVgqBJnZx1uHUyYcUxMVeZz52eYcTUMReH63ghTlVltmqR5yeEk5tyMw1Ld4uycgyyqKfnv3OkSJDmaIhOkOYeTlJfWGsiyhEwlTHLn0EcksEJlOHH/7+STmpY8iPcqi/nj4idZXTrFRwMfu6D8XmTyPop4Wrv5py3a/yhFsPd7fSEEpRAM/KQqOU8iCiGRF5XN43LTQpblRz6Ae37CaztjdgYhqiozCjOatspa0+Tl7TGaIjGOc6wpP/pTZ1q0pr3n1jQrF0Jg6+o75XRJ5hefm2fvoMtyrOKaKp4uM0kFC4382OfYNlRWWhb/endwwjJzvVVtulxDpmZpFEXJwnoDXZGRgKwoCZKCM7Mue+OYcZwRpwWfOdNCV2Uuz7v8wtV54qzA1BS2+yFvHkzoBjF5KfDTnCu2i7XkIikyl+drXF1weWG1ydagKr2XQmKhbnJzzycvS87NVp7Jr26NuLZcP86Ot/eih3TP4aSO9eNMSx6Fj/vv+yl++vCxC8o/Sf7thwUf9G7+aWcnD167H6Tc3Btz6KfHfeYnvf7Rve8MQmRJ4vsbfVSlou54psZ37/TwDIUXVpuPPFdnkrAzCMlKwdv7E5Ks5Ic7Cf/pmQVajkbd1nhxtc6sZ2LpKnM186HzPG7jlAwOEYZ1/H2eAZ6hcnbmHdnJlqOzUKsGsu6XxdwaRLy1H+CZGp1JRBxWlpbPLte4e+iTCwldFnz+wgzbw4irix5tR2e5aZ8UxghS9ocRWS4wVBlVkVhvOWz2IlxLw5YlmraOpanHWuMzns7uMKYoQZYlGqaOoVYl9KMpd1Wv+vYCiNKTylhwssLzONOSR+H09/0UP2342AXlj1JP9cOKDyI7eS8OPhu9/7+9e41tq777AP49x5fjay52eoGkDhToWkG7Nn3B9EwDIUAMhoQEanphrQZ7U6RJo6Ctk6ZNnVp12lSkaRPTNnUriDcMtr1hj7SpGmioG+oDLZTL07Q8GS1paZM0thNf4uPL+T8vTs6J7RwnTuLYx/b386aNnRP//3acn/+//+WXxsdXJ80V2b0zf8yrefzZtktwuyTc3OXFf8ZTWN2hIORzweV04OJoEt1+xTIgaELA6ZTxf9cTEAJwOwBNk/DRlUn4FSemMnl09LoQ8Dig790twOrMOqsPTh65gL6u+acejBF38ajbWCB1OTYNt6zPQU+lc/C5Xchl9cVdEvRDQBRZwqa1QdzZ24n+mTR9+cEYNxIqCkLD+tUB3NKjp7ZvpHPo9Ltw580dcDskXIlnsHrmrPFbw35oGjCVySGWzsHt0D8sdEJZ2MUAABfbSURBVHpdmEhm4XHK5j7zGxMphK/rR/AWf4haaoaH73dqNm33m9mqc6p2UuvRyWIr+BhlIoHSxUDVPH4qkzPrDrscDmiaBqdDhlOWkNeA/1xPIJ0tQAh9RFfehrUdHv1c75kGZAtAp8eJmzo98Lj0ylL5XAEJNY9cXiCR8UDTNMiyXKFFpRYbmMoXSMkOGe9/Pok1HR54hQSnS0ZfyAuhCSSzBVybzMDr0rd7eWfKbKazBWhCYCSW1s+pdjshSTLSakEv1+h24OauLLb0dcLtkM39wkZZznUhPyBJSKl59PgV5Aoabg3rt21d14WCpplzxOGAgoDiqLhNb7EZHr7fqdm0XVAu/sTtzcbMM4jbYZFXvdR6dLLYCj7FK3QBmFtrFnp8IQRGE6o5ynZKwM2dPnwxmUFAceHSRAo9fgUOCXC7rANHt9+NbZFufDGpIlvQkM9rUJx6FaZ1XT5cjaeRyWv473PXsCrowUg0jWvxDO7buMoMzPNlBRYbmMoXSF2JpSEEkC9o6On2wq84EVSAsN+Nc1cm0dvlQZfPDQiB//nPBFZ3epEraChoApPpLHpDPgQUB3oCbtxIZiFLMrr9HgQVd0lABqzLcvrc+mtjHALT5XVi6HoC0XQOisuB+HjC3NZWixQzV0RTs2m7oAwU/WFTBOeVVkCtRyfzjby7fS6zYLwmBCSIOSt0FZfD3OtqblUqWi1tBEBNCMRSKsIzAScvgMnpLB6+ay0Sqr6S2iHNjuaMNpSP5r68rgsFAQyPJVGYqZncF/Kj2++ERwnijXNf4KYuLzq9+uEbH4zEcfuaAPrD/nmzAuUWc1yqsUAq6NF/TijgxtoOD4zxeTSVxXhSPwxlIpVDQHFgdDKDaDqHggC8LhkTiZmsg2e2/OWGNUGsDioYm8rgSpVlOQGg+ADdtZ1efDFzbXEVp1qlmIsfv94LEIkWqy2DMq2sWo9OKo683Q5cuJ4wC8YbBQAAYa7QDXqc2Lg2iK6ZGsDFwa63ywMJMINJOptHNJVFb5fHrLesuBy4q7dTHwEKmCdJGaM5q7bJsozt/d1YvyqApJpDNKkimlSR0yTEUipuCfnQ4XNCzc3uJZ5IqugP++fNChSrJqUvhNCrGrlkjCWzCCoykpk8hNAXpN1IZBEJ+XDHGh9Gp7Il2YXRyQzGkirCQQWFnIZsXl95PjWd02tNQ597Nh6v2+/G6iq3n5WrV4q5kQsQiarFoEwropYrvCv90RbAzGrcmcesUABAkvRDYsqD3dBoEj6XAw559hSwiaI5aOOITuOoyMUc21jc/74uL85+HtdHzpqG2HQWI/FpyDOrj28J+RCeeZ7KswICAkm1gEs3kpA0B4QQkCRpwZR+cQCSIOBzOeByyHA4Ctjc14FkVoOaK8DplKG4nHDKuZLsgjbTP+Nc74IAHLKEzX36yvHy6kjLGX3Wa0qJ26OoGTAok+1VGnmX/4EFKhcAsEqBq7kCHJJe0g+YnXfN5gpAWXm/OW1wO8wPBQsFqOIqVIWCE+9djmMsoaKvy6uPZN0OpLJ5RJMqNE1DOpuH4nTAIQtcjWUwkczC53Lg2hdT8IYS+NLaYEl/jMCt5gq4PjldcoKVfr++hWhyOgdIEiRJRlCRzQ8dQszd/+tySPginkYsPbs9KeR345awH91+t/nzjXrU5SenLXb0WY8pJW6PombAoExNwWrkvZgFZVa3KS79mEmDEZhuLxppd3mdc4Jst89lmQbdsCYwJ0Xe1+2F3z37GGpBoD/kRZfPhVV+F3weFyZTWZy7Mol3czF43Q5kcgWMJ7JYFVQQT+kpZqMOszGym50rFmb9YQlAt88NWZLM87pL+ut0IJ7Omh84DGs7PJAlvYqWY6bucF+XB/1hHy6Mpsw0/sY1AXSV9b2gCVyfnDbrHQP2HX1yexQ1A/42UtMy09pXZm+rlFK2SoFvXBMomVMGgN5un7n/2EgBj8TSSKoFZHMF3LY6gFvCPss0qNftsLz9jjUB82vF6YAkyej0urGux4//jKUgJH0LlXEK1129HQh63RifmsaXbtL3/Ravaja2GvV1e3H+esIMyEbwvhpL46ZODxKZHBwy4HHqKXqnDNy2OoB4erakYV+3F91+tz4nbHFM6dpOX8lt5SlgNV/AeDKLDp9eoat8xF6rFHQtFmhxexQ1AwZlalpGSnk6unDB+EopcAAVFyjF0jmMxNLmSBTQg2yyvwsSREmgBPQjNq3bCTMYGCnydDZvFm7oCbjNVdD6CVeanlru0LcjuRylb9PirUYA4JD0YO+c+SGxVBbnr03hi3gGU2oet4R8uDXsx529HdiwNoj4dN6yv1ZrAMpvm1N7emb7mZrNYzKdnTNir8Uiqlot0OL2KGoGDMrU1CRJmrdgvNUIa6HAY0ipeSTVghmQAT1oXp9S0e1zm/V7DeGAgmgqh3IBxYV13b45e3XHEio0IZkjTABmAQn9Or2u841kzryvfKvR6g5P0dyxnk7+bELfCx0KuOFVnEioechO/XtlWV7WArzyVK/xISOvaXNG7LVKY9dygRYLRpDdMShTw6zknlFN0zASTeOTa1OYTOfQE3BDkuRFjbD8ilNf9FVEgoDi1AsfJ9U8vC4H1LyGSMiLvi5PxRXaVsGge+b/V2LTZgD2uZ3m/PG6bh82rAmYI9uw6JjT9vKUrJrXi0bo3yLpp3K5HHBIElLZQsn+4KUofzwBCZt7OyAkwK8kzRF78aleyy0SwgVa1E4YlKkhylOSEgTCATdCAQUBxbWsAK1pGt4aGsfQ9Smcv54ABHBrjx9b+joWNcLq9rlw2+qAue1KgoDf48TlG2nceXMQak7Dleg0eru9GE9m8elYChvWBKpOj5anU7dHuiGgH2tpjKavxjPwK070dXuRHC0sWLxCEwIFTWB4PDX7PdBH37VY0FQpBRxL53A1ps/Ni6LvNxekLSMFzQVa1E74W00NUZySlCAQTeXwwcgk1q8OIDgThJY6HzkSm8YHI3F9znYmQnx2I4WbujxYHfRUPcKSJAkDkS44ZeDz6DQKQuDyeAq93T4ISLN7pCUJ0szXRsCvdgRnedpVhQBWzc8QQiCt5jE1nbMs31gLlqP+BRZRLScFzQVa1E4YlKkhilOSeQ3moRXGOdXLmY+cSKoQAFxOGZJk7MMFEpkcVgc9ixphybKML6/rRiQcwKUbSfhceno5lZ0temG02ejXSs2hhsXcbU7lJEnCxps6sKbDU1K+sbja00pMGSy0iKqaFLQQwjwytfh6LtCidsKgTA1RHBiLqzoprtnAs9QAFw4okADk8hr6Qz5cjqYBAXR4XEsaYRkjQyCAaCqmn6ddVPSiuM3F6dqlBr9KASyd06pvb1H5RkOlFHLxvPVyAt58i6gWSkEbbTOOTDXaZmRLuECL2gWDMjVEcUrSCHDhgNss9AAsfc5wXbcXW9d14YOROJwz9YEjIR/+6/YwQn5lySOs4jYXb20y2mwE/OVu4anUb5+ruvKOlViNwK/G0kiq+Tl7l2t9HnS16e3i+Wi7HkJCtJIYlKkhSlKSmRzCATdi6ax5KtRy5gxlWcZ9G1fh9jUBTCRVhAMK1nV7q65ZXFWbLcoQznfO9mIXmFme9R2PLqv9ViPwvKZXswoXjapXIhjWIr1N1A4YlKlhSoo2hHw1neuUZRn9YT/6w/4atnjhMoTA8gNMpQA2FF/w0hLlKXSfe+6ctJovwO2ae3sqkzPbXKs53OWkt4naBX/jyRZaac6wFgGmmudjvnlrqxR6b5cHfV2ekmNFIyGvWUfZfGwIjCZUxEeT5m0rXeJwMUemErUyBmWiGlvuFp5qFoktNG9tOX8cz2Ag0lVyrGiX1zmniEY44J7ZTjX7mCs9v7uYI1MNK3n4DFGjMCgT1dhytvDMF2yLLbTvt2IKPasfSVocXMvbmlRz5tGeJdeu8PzuQkemFqvVedhEdrO8lS9LcPLkSTz//POW97322mt4/PHHMTg4iLfeeqvOLSOqHSP9bATAagNFpWAbS5cGyfnmrYHFpdDL2xpQrEf0frcD0VQWI9E0oqkshBCW31cP1T5PRM2mriPlI0eO4NSpU9i0adOc+8bHx/HKK6/gz3/+M1RVxZ49e/DVr34VbnfzzzFSe6hFOnWhYGtYKOgaKfSrsTTymr6gKxLyosu78FveKv3e2+XB2FSmZD66kSNTrtamVlXXkfLAwAAOHTpked+HH36Ibdu2we12IxgMIhKJYGhoqJ7NI1oyI5169nKs5N/FjiarHeEagbNY+bx1T8ANQMK1eBoFTWA8mcXF0eSCbTLS7wP93ea/qztKF4gBjR2ZcrU2taoV+Q1+/fXX8fLLL5fcdvToUTzyyCM4ffq05TXJZBLB4Oy8md/vRzKZtPxeg6qqOH/+/JLbmclklnV9M2Af69QGzYFzV6dKD7+4AkxHO+CRCxWvs5QWuDJz7KgEoC/kw+jnUct+hoUD6ZwGn0uGiEfNbVNXUwLXEjn87/UkIPRTzsIe4NLnMq6PBtDpQtXtSgKIqhJGxua+H73ZGEJK7dLYi3otKzxPozVrzcqxw+8s2dOKBOUdO3Zgx44di7omEAgglZqtbJNKpUqCtBVFUSxT4dU6f/78sq5vBuxjfYxE0+bxkMV61garWrhUbGOFNHi1/Yymsrh6OYZOZx7hrAeAvs1JdTuRVAuYgh95yYm+rurTz9FUFtPu2JzbN/V31zRdvJjXstLz1Awa/TvLDwT2VfeFXpVs2bIFZ86cgaqqSCQSGB4exoYNGxrdLKKq1DKdutRFYgZjvtU4vhQACgK4NpkxyzgCldPPQog5C7qqSZfX23KfJyI7avgEzIkTJxCJRHD//fdj79692LNnD4QQOHDgABRFWfgHENmAncoLGh8EjPO5P4+mkStocDlkREI+OOXZmsflC6Pm22rESk1EK6/uQfnuu+/G3XffbX791FNPmf8fHBzE4OBgvZtEtGx2Ki9Y/AEh5Hehw9sJxQmoeQGnLEEUHQpSPpJfaP9zq5y6RmRXDR8pE7UKuxwVavUBwerkLquRPLcaETUWgzJRC7L6gFDNSJ5bjYgai+80ojZRzUjeTnPjRO2IQZnIpspPCKsHO82NE7UjBmUiG7JaBY20wEYhVjxA2mVunKgd2WafMhHNslwFHU2z4AJRi2NQJrIhq1XQosLtRNQ6GJSJbMiyxGKF24modTAoE9mQ5bGWIR9XQRO1OH7sJrIhq1XQo59HuQqaqMUxKBPZVPkq6GYoSUhEy8OgTNSEyvcwcy8xUWtgUCZqMvNVcmJgJmpuXOhFLc2qNnCzq1TJaSX3MLfi80hkRxwpU8tq1RFlvSs5terzSGRHHClTy2rEiLIe6l3JqVWfRyI7YlCmljXfiLKZWe5hXsFKTq36PBLZEdPX1LJatTZwvSs5terzSGRHHClTy6r3iLKejD3M60I+hPzuFZ3bbeXnkchu+FGXWlaz1wZuRD1lK83+PBI1EwZlamnNWhu4kfWUrTTr80jUbJi+JrIh1lMmak8MykQ2ZMd6yjxAhGjlMX1NZEN2q6fMA0SI6oMjZSIbsls9ZR4gQlQfHCkT2ZDd6inX+2hPonbFoExkU3aqp8wDRIjqg+lrIloQDxAhqo+6f8w9efIk/va3v+GFF16Yc9+RI0dw9uxZ+P1+AMCvf/1rBIPBejeRiMrwABGi+qhrUD5y5AhOnTqFTZs2Wd7/ySef4Pjx4wiFQvVsFhFVgQeIEK28uqavBwYGcOjQIcv7NE3D5cuX8eMf/xi7du3Cn/70p3o2jYiIqOFWZKT8+uuv4+WXXy657ejRo3jkkUdw+vRpy2vS6TS++c1v4qmnnkKhUMC+fftw1113YePGjRUfR1VVnD9/fsntzGQyy7q+GbCPraMd+tkOfQTap5+0eCsSlHfs2IEdO3Ys6hqv14t9+/bB69UXk3zlK1/B0NDQvEFZUZSKqfBqnD9/flnXNwP2sXW0Qz/boY9A4/vJDwT2ZZvV15cuXcLu3btRKBSQy+Vw9uxZ3HnnnY1uFhERUd00fJPhiRMnEIlEcP/99+Oxxx7D4OAgXC4XHnvsMdxxxx2Nbh4REVHdSKKJT5X/4IMPoChKo5tBRNRUVFXF1q1bG90MstDUQZmIiKiV2GZOmYiIqN0xKBMREdkEgzIREZFNMCgTERHZBIMyERGRTTAoExER2UTDDw+pt3YpHTlfP1977TW8+uqrcDqdeOaZZ3Dfffc1oIVLl8lk8L3vfQ8TExPw+/342c9+Nqey2DPPPINYLAaXywVFUXD8+PEGtXZxNE3DoUOHcOHCBbjdbhw5cgT9/f3m/c3+2hkW6mcrvRfPnTuHY8eO4ZVXXim5/c0338SLL74Ip9OJJ554AoODgw1qIdmKaCOHDx8WDz30kHj22Wct79+1a5eYmJioc6tqb75+jo2NiUcffVSoqiqmpqbM/zeTP/zhD+KXv/ylEEKIv/71r+Lw4cNzvufhhx8WmqbVu2nL9ve//10cPHhQCCHE+++/L/bv32/e1wqvnWG+fgrROu/F3/3ud+LRRx8VO3bsKLk9m82KBx54QMTjcaGqqnj88cfF+Ph4g1pJdtJW6et2KR05Xz8//PBDbNu2DW63G8FgEJFIBENDQ/Vt4DKdOXMGX/va1wAA99xzD955552S+2/cuIGpqSns378fu3fvxltvvdWIZi5Jcd+2bt2Kjz/+2LyvFV47w3z9bKX3YiQSwa9+9as5tw8PDyMSiaCzsxNutxvbt2/Hu+++24AWkt20ZPq6XqUjG20p/UwmkyVpQL/fj2QyuaLtXA6rPobDYbMPfr8fiUSi5P5cLoenn34a+/btw+TkJHbv3o0tW7YgHA7Xrd1LlUwmEQgEzK8dDgfy+TycTmfTvXbzma+fzfherOShhx7ClStX5tzeSq8l1VZLBuV6lY5stKX0MxAIIJVKmV+nUilbz9VZ9fE73/mO2YdUKoWOjo6S+3t6erBr1y44nU6Ew2Fs2rQJn332WVME5fLXR9M0OJ1Oy/vs/trNZ75+NuN7cbFa6bWk2mqr9PV82qV05JYtW3DmzBmoqopEIoHh4WFs2LCh0c1alIGBAfzzn/8EALz99tvYvn17yf3//ve/8d3vfheA/sfu008/xfr16+vezqUYGBjA22+/DUAvuFL82rTCa2eYr5/t8F687bbbcPnyZcTjcWSzWbz33nvYtm1bo5tFNtCSI+XFaJfSkcX93Lt3L/bs2QMhBA4cONB0lbZ2796NgwcPYvfu3XC5XOYK85///Of4+te/jnvvvRenTp3C4OAgZFnGc889N2d1tl09+OCD+Ne//oVdu3ZBCIGjR4+21GtnWKifrfpefOONN5BOp7Fz50784Ac/wLe//W0IIfDEE09gzZo1jW4e2QCrRBEREdkE09dEREQ2waBMRERkEwzKRERENsGgTEREZBMMykRERDbBoExERGQTDMpEZf7yl7/g2LFjlvfF43G88cYbi/6Z1V43PT2NXbt2YXh4eNGPQUTNj0GZaBEuXLiAN998c0Wu++ijj/Dkk09iZGRkqc0joibX9id6EVXywgsv4OOPP0Y8HsfGjRvx05/+FL/5zW8wNDSEP/7xj7jnnnvwox/9CKqqQlEUHD58GIVCAc8//zzWrl2LkZERbN68GT/5yU9Krtu5c6fl42WzWbz44ov4/ve/X+eeEpFdMCgTWcjlcujp6cGJEyegaRq+8Y1vYHR0FPv378err76KnTt34tlnn8XevXtx77334p133sGxY8dw4MABXLp0Cb///e/h9XrxwAMPYHx8vOS6SsrP8Cai9sOgTGRBkiREo1E899xz8Pl8SKfTyOVyJd9z8eJF/Pa3v8Xx48chhDCrHEUiEbMs4apVq6Cqat3bT0TNiUGZyMLp06fR39+PX/ziF4hGozh58iSEEJBlGZqmAQDWr1+Pp59+GgMDAxgeHjaL1EuSNOfnFV9HRFQJgzKRhc2bN+OTTz7Bk08+CUmSsG7dOoyNjSESieDixYt46aWXcPDgQRw6dAiqqiKTyeCHP/xhxZ9XfN23vvWt+nWEiJoKq0QRERHZBEfKRHX0j3/8Ay+99NKc2/ft24cHH3yw/g0iIlvhSJmIiMgmeHgIERGRTTAoExER2QSDMhERkU0wKBMREdnE/wM9esFGG0jBTwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.scatterplot(\n",
    "    x=\"latent_1\", y=\"latent_2\",\n",
    "    #hue=\"label\",\n",
    "    #palette=sns.color_palette(\"hls\", latent_movie[\"rating\"].nunique()),\n",
    "    data=inner_raw_item,\n",
    "    legend=\"full\",\n",
    "    alpha=0.3,\n",
    ")\n",
    "\n",
    "# Test if the lord of the ring movies are close together\n",
    "test_viz = inner_raw_item[inner_raw_item[\"title\"].str.contains(\"Lord of the R\")].reset_index()\n",
    "\n",
    "for i, titl in enumerate (test_viz['title']):\n",
    "    plt.text(test_viz['latent_1'][i]+0.2, test_viz['latent_2'][i]+0.2, titl)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "** Optional: do the same for the users and check which users are close together **"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### That was the hands on, good job! Do you have time left? There is plenty to explore learn with this dataset on your own. Some ideas:\n",
    "\n",
    "5.1 Optional: create an ensemble of the two different models (or any other) and see how that one performs on the test set. Does it improve performance (in RMSE or confusion matrix statistics)?\n",
    "\n",
    "5.2 Optional: Try out the EASE model: https://github.com/Darel13712/ease_rec  \n",
    "\n",
    "5.3 Optional: Add yourself as a user, add some ratings of movies that you have seen, and recommend movies for yourself"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}