Notebook X. Summarizing Classification Outcomes.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Notebook X. Summarizing Classification Outcomes.ipynb","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"}},"cells":[{"cell_type":"markdown","metadata":{"id":"v122wsZWXYyR"},"source":["# **Notebook X.** Summarizing Classification Outcomes\n","----\n"]},{"cell_type":"markdown","metadata":{"id":"MBgbXYClaFio"},"source":["In  the earlier notebooks we have tested different classification models. \n","\n","Here, we provide a comparison of all of the models as well as an in-depth analysis of the \"best performing\" model. "]},{"cell_type":"markdown","metadata":{"id":"z7pNU-UvaR-1"},"source":["# X.1. Preamble: Load Packages \n","---"]},{"cell_type":"code","metadata":{"id":"Jcub0qN1SvYt"},"source":["# General Packages #\n","import os\n","import pandas as pd\n","import numpy as np\n","\n","# Load TQDM to Show Progress Bars #\n","from tqdm import tqdm\n","from tqdm.notebook import tqdm as tqdm_notebook\n","\n","# Sklearn Packages #\n","from sklearn.feature_extraction.text import TfidfTransformer, CountVectorizer, TfidfVectorizer\n","from sklearn.metrics import mean_squared_error, r2_score, accuracy_score, classification_report\n","\n","from sklearn.metrics import accuracy_score, roc_auc_score, precision_score, recall_score, f1_score, confusion_matrix\n","\n","import matplotlib.pyplot as plt\n","import seaborn as sns"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"V1iD6MGr2ZS4"},"source":["# Turn of warnings, just to avoid pesky messages that might cause confusion here\n","# Remove when testing your own code #\n","import warnings\n","warnings.filterwarnings(\"ignore\")"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"EpKLFW99S1Sz","executionInfo":{"status":"ok","timestamp":1627938893135,"user_tz":420,"elapsed":21318,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"82184fb4-a799-4b82-d60d-b18f815027c2"},"source":["# Mount Personal Google Drive on own Machine -- You have to follow the link to log in #\n","from google.colab import drive\n","drive.mount('/content/drive')"],"execution_count":null,"outputs":[{"output_type":"stream","text":["Mounted at /content/drive\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"O-8BdJPq1rrS"},"source":["# X.2. Compare Classification Results ##\n","---------------- \n","\n","We have saved our model classification results in Google Drive. Here, we are going to compare the different classification results. "]},{"cell_type":"code","metadata":{"id":"8MzdWDF12jYI"},"source":["# Change to Working Directory with Training Data # \n","os.chdir(\"/content/drive/MyDrive/USPTO AI Patent Classification/\")\n","\n","# Collect a list of all of the files where we saved classification results # \n","\n","Counter = 0\n","for root, dirs, files in os.walk(\"./Output/Model Performance/\"):\n","  for file in files: \n","    temp = pd.read_csv(\"./Output/Model Performance/\" + file)\n","    if Counter == 0:\n","      Combined = temp\n","    else:\n","      Combined = pd.concat([Combined, temp])\n","    Counter = Counter + 1\n"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"PIrtNzAY7yqh","executionInfo":{"status":"ok","timestamp":1627938899852,"user_tz":420,"elapsed":8,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"bcb38a41-fa43-4983-b161-08ea93fb9acf"},"source":["# See the Ouptut with All Models #\n","\n","Combined.drop(\"Unnamed: 0\", axis = 1).sort_values(\"Accuracy\", ascending = False)"],"execution_count":null,"outputs":[{"output_type":"execute_result","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>Name</th>\n","      <th>Type</th>\n","      <th>Share</th>\n","      <th>True-Positives</th>\n","      <th>False-Negatives</th>\n","      <th>False-Positives</th>\n","      <th>True-Negatives</th>\n","      <th>Accuracy</th>\n","      <th>AUC</th>\n","      <th>Precision</th>\n","      <th>Recall</th>\n","      <th>F1</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>Random Forest</td>\n","      <td>Bag of Words</td>\n","      <td>0.183</td>\n","      <td>0.940</td>\n","      <td>0.965</td>\n","      <td>0.060</td>\n","      <td>0.035</td>\n","      <td>0.960</td>\n","      <td>0.922</td>\n","      <td>0.940</td>\n","      <td>0.858</td>\n","      <td>0.897</td>\n","    </tr>\n","    <tr>\n","      <th>0</th>\n","      <td>SciBERT</td>\n","      <td>Transformer</td>\n","      <td>0.201</td>\n","      <td>0.882</td>\n","      <td>0.970</td>\n","      <td>0.118</td>\n","      <td>0.030</td>\n","      <td>0.952</td>\n","      <td>0.926</td>\n","      <td>0.882</td>\n","      <td>0.882</td>\n","      <td>0.882</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>PatentSBERTa</td>\n","      <td>Transformer</td>\n","      <td>0.211</td>\n","      <td>0.858</td>\n","      <td>0.975</td>\n","      <td>0.142</td>\n","      <td>0.025</td>\n","      <td>0.950</td>\n","      <td>0.932</td>\n","      <td>0.858</td>\n","      <td>0.901</td>\n","      <td>0.879</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>BioBERT</td>\n","      <td>Transformer</td>\n","      <td>0.211</td>\n","      <td>0.858</td>\n","      <td>0.975</td>\n","      <td>0.142</td>\n","      <td>0.025</td>\n","      <td>0.950</td>\n","      <td>0.932</td>\n","      <td>0.858</td>\n","      <td>0.901</td>\n","      <td>0.879</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>ALBERT</td>\n","      <td>Transformer</td>\n","      <td>0.194</td>\n","      <td>0.890</td>\n","      <td>0.964</td>\n","      <td>0.110</td>\n","      <td>0.036</td>\n","      <td>0.950</td>\n","      <td>0.915</td>\n","      <td>0.890</td>\n","      <td>0.857</td>\n","      <td>0.873</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>DistilBERT</td>\n","      <td>Transformer</td>\n","      <td>0.219</td>\n","      <td>0.834</td>\n","      <td>0.976</td>\n","      <td>0.166</td>\n","      <td>0.024</td>\n","      <td>0.945</td>\n","      <td>0.931</td>\n","      <td>0.834</td>\n","      <td>0.907</td>\n","      <td>0.869</td>\n","    </tr>\n","    <tr>\n","      <th>5</th>\n","      <td>BERT</td>\n","      <td>Transformer</td>\n","      <td>0.215</td>\n","      <td>0.837</td>\n","      <td>0.973</td>\n","      <td>0.163</td>\n","      <td>0.027</td>\n","      <td>0.944</td>\n","      <td>0.925</td>\n","      <td>0.837</td>\n","      <td>0.894</td>\n","      <td>0.865</td>\n","    </tr>\n","    <tr>\n","      <th>6</th>\n","      <td>XLNet</td>\n","      <td>Transformer</td>\n","      <td>0.218</td>\n","      <td>0.833</td>\n","      <td>0.974</td>\n","      <td>0.167</td>\n","      <td>0.026</td>\n","      <td>0.944</td>\n","      <td>0.928</td>\n","      <td>0.833</td>\n","      <td>0.901</td>\n","      <td>0.866</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>Support Vector Classifier (RBF)</td>\n","      <td>Bag of Words</td>\n","      <td>0.174</td>\n","      <td>0.912</td>\n","      <td>0.949</td>\n","      <td>0.088</td>\n","      <td>0.051</td>\n","      <td>0.943</td>\n","      <td>0.886</td>\n","      <td>0.912</td>\n","      <td>0.791</td>\n","      <td>0.847</td>\n","    </tr>\n","    <tr>\n","      <th>7</th>\n","      <td>RoBERTa</td>\n","      <td>Transformer</td>\n","      <td>0.224</td>\n","      <td>0.821</td>\n","      <td>0.977</td>\n","      <td>0.179</td>\n","      <td>0.023</td>\n","      <td>0.942</td>\n","      <td>0.931</td>\n","      <td>0.821</td>\n","      <td>0.913</td>\n","      <td>0.865</td>\n","    </tr>\n","    <tr>\n","      <th>8</th>\n","      <td>ELECTRA</td>\n","      <td>Transformer</td>\n","      <td>0.204</td>\n","      <td>0.840</td>\n","      <td>0.962</td>\n","      <td>0.160</td>\n","      <td>0.038</td>\n","      <td>0.938</td>\n","      <td>0.905</td>\n","      <td>0.840</td>\n","      <td>0.851</td>\n","      <td>0.846</td>\n","    </tr>\n","    <tr>\n","      <th>0</th>\n","      <td>Doc2Vec USPTO Patent Embeddings</td>\n","      <td>CNN</td>\n","      <td>0.203</td>\n","      <td>0.842</td>\n","      <td>0.963</td>\n","      <td>0.158</td>\n","      <td>0.037</td>\n","      <td>0.938</td>\n","      <td>0.906</td>\n","      <td>0.842</td>\n","      <td>0.853</td>\n","      <td>0.848</td>\n","    </tr>\n","    <tr>\n","      <th>9</th>\n","      <td>Big-Bird</td>\n","      <td>Transformer</td>\n","      <td>0.226</td>\n","      <td>0.801</td>\n","      <td>0.974</td>\n","      <td>0.199</td>\n","      <td>0.026</td>\n","      <td>0.935</td>\n","      <td>0.922</td>\n","      <td>0.801</td>\n","      <td>0.901</td>\n","      <td>0.848</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>GLOVE (6B - 50)</td>\n","      <td>CNN</td>\n","      <td>0.190</td>\n","      <td>0.855</td>\n","      <td>0.953</td>\n","      <td>0.145</td>\n","      <td>0.047</td>\n","      <td>0.934</td>\n","      <td>0.888</td>\n","      <td>0.855</td>\n","      <td>0.810</td>\n","      <td>0.832</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>Logistic Regression</td>\n","      <td>Bag of Words</td>\n","      <td>0.225</td>\n","      <td>0.799</td>\n","      <td>0.973</td>\n","      <td>0.201</td>\n","      <td>0.027</td>\n","      <td>0.934</td>\n","      <td>0.919</td>\n","      <td>0.799</td>\n","      <td>0.895</td>\n","      <td>0.844</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>GLOVE (840B - 300)</td>\n","      <td>CNN</td>\n","      <td>0.194</td>\n","      <td>0.834</td>\n","      <td>0.952</td>\n","      <td>0.166</td>\n","      <td>0.048</td>\n","      <td>0.929</td>\n","      <td>0.884</td>\n","      <td>0.834</td>\n","      <td>0.808</td>\n","      <td>0.821</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>Gradient Boosting Classifier</td>\n","      <td>Bag of Words</td>\n","      <td>0.224</td>\n","      <td>0.782</td>\n","      <td>0.967</td>\n","      <td>0.218</td>\n","      <td>0.033</td>\n","      <td>0.926</td>\n","      <td>0.905</td>\n","      <td>0.782</td>\n","      <td>0.872</td>\n","      <td>0.824</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>Multi Layer Perceptron</td>\n","      <td>Bag of Words</td>\n","      <td>0.201</td>\n","      <td>0.809</td>\n","      <td>0.953</td>\n","      <td>0.191</td>\n","      <td>0.047</td>\n","      <td>0.924</td>\n","      <td>0.882</td>\n","      <td>0.809</td>\n","      <td>0.812</td>\n","      <td>0.810</td>\n","    </tr>\n","    <tr>\n","      <th>5</th>\n","      <td>Ridge Regression</td>\n","      <td>Bag of Words</td>\n","      <td>0.219</td>\n","      <td>0.779</td>\n","      <td>0.962</td>\n","      <td>0.221</td>\n","      <td>0.038</td>\n","      <td>0.922</td>\n","      <td>0.895</td>\n","      <td>0.779</td>\n","      <td>0.852</td>\n","      <td>0.814</td>\n","    </tr>\n","    <tr>\n","      <th>6</th>\n","      <td>Decision Tree with Boosting</td>\n","      <td>Bag of Words</td>\n","      <td>0.234</td>\n","      <td>0.759</td>\n","      <td>0.970</td>\n","      <td>0.241</td>\n","      <td>0.030</td>\n","      <td>0.920</td>\n","      <td>0.907</td>\n","      <td>0.759</td>\n","      <td>0.884</td>\n","      <td>0.817</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>FastText</td>\n","      <td>CNN</td>\n","      <td>0.193</td>\n","      <td>0.811</td>\n","      <td>0.946</td>\n","      <td>0.189</td>\n","      <td>0.054</td>\n","      <td>0.920</td>\n","      <td>0.868</td>\n","      <td>0.811</td>\n","      <td>0.782</td>\n","      <td>0.796</td>\n","    </tr>\n","    <tr>\n","      <th>7</th>\n","      <td>LASSO</td>\n","      <td>Bag of Words</td>\n","      <td>0.221</td>\n","      <td>0.771</td>\n","      <td>0.961</td>\n","      <td>0.229</td>\n","      <td>0.039</td>\n","      <td>0.920</td>\n","      <td>0.894</td>\n","      <td>0.771</td>\n","      <td>0.850</td>\n","      <td>0.809</td>\n","    </tr>\n","    <tr>\n","      <th>0</th>\n","      <td>Multi Layer Perceptron</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.213</td>\n","      <td>0.761</td>\n","      <td>0.951</td>\n","      <td>0.239</td>\n","      <td>0.049</td>\n","      <td>0.911</td>\n","      <td>0.873</td>\n","      <td>0.761</td>\n","      <td>0.809</td>\n","      <td>0.784</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>No Embeddings</td>\n","      <td>CNN</td>\n","      <td>0.193</td>\n","      <td>0.777</td>\n","      <td>0.938</td>\n","      <td>0.223</td>\n","      <td>0.062</td>\n","      <td>0.907</td>\n","      <td>0.848</td>\n","      <td>0.777</td>\n","      <td>0.749</td>\n","      <td>0.763</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>Random Forest</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.168</td>\n","      <td>0.815</td>\n","      <td>0.924</td>\n","      <td>0.185</td>\n","      <td>0.076</td>\n","      <td>0.906</td>\n","      <td>0.823</td>\n","      <td>0.815</td>\n","      <td>0.685</td>\n","      <td>0.744</td>\n","    </tr>\n","    <tr>\n","      <th>8</th>\n","      <td>Decision Tree with Bagging</td>\n","      <td>Bag of Words</td>\n","      <td>0.213</td>\n","      <td>0.746</td>\n","      <td>0.947</td>\n","      <td>0.254</td>\n","      <td>0.053</td>\n","      <td>0.904</td>\n","      <td>0.863</td>\n","      <td>0.746</td>\n","      <td>0.793</td>\n","      <td>0.769</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>Gradient Boosting Classifier</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.240</td>\n","      <td>0.699</td>\n","      <td>0.957</td>\n","      <td>0.301</td>\n","      <td>0.043</td>\n","      <td>0.895</td>\n","      <td>0.873</td>\n","      <td>0.699</td>\n","      <td>0.837</td>\n","      <td>0.762</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>Logistic Regression</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.271</td>\n","      <td>0.670</td>\n","      <td>0.974</td>\n","      <td>0.330</td>\n","      <td>0.026</td>\n","      <td>0.892</td>\n","      <td>0.897</td>\n","      <td>0.670</td>\n","      <td>0.906</td>\n","      <td>0.771</td>\n","    </tr>\n","    <tr>\n","      <th>5</th>\n","      <td>LASSO</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.268</td>\n","      <td>0.668</td>\n","      <td>0.970</td>\n","      <td>0.332</td>\n","      <td>0.030</td>\n","      <td>0.890</td>\n","      <td>0.890</td>\n","      <td>0.668</td>\n","      <td>0.892</td>\n","      <td>0.764</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>Ridge Regression</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.269</td>\n","      <td>0.668</td>\n","      <td>0.971</td>\n","      <td>0.332</td>\n","      <td>0.029</td>\n","      <td>0.890</td>\n","      <td>0.892</td>\n","      <td>0.668</td>\n","      <td>0.895</td>\n","      <td>0.765</td>\n","    </tr>\n","    <tr>\n","      <th>9</th>\n","      <td>Robust Regression</td>\n","      <td>Bag of Words</td>\n","      <td>0.241</td>\n","      <td>0.688</td>\n","      <td>0.954</td>\n","      <td>0.312</td>\n","      <td>0.046</td>\n","      <td>0.890</td>\n","      <td>0.866</td>\n","      <td>0.688</td>\n","      <td>0.827</td>\n","      <td>0.751</td>\n","    </tr>\n","    <tr>\n","      <th>10</th>\n","      <td>Linear Regression</td>\n","      <td>Bag of Words</td>\n","      <td>0.232</td>\n","      <td>0.693</td>\n","      <td>0.948</td>\n","      <td>0.307</td>\n","      <td>0.052</td>\n","      <td>0.889</td>\n","      <td>0.856</td>\n","      <td>0.693</td>\n","      <td>0.800</td>\n","      <td>0.743</td>\n","    </tr>\n","    <tr>\n","      <th>6</th>\n","      <td>Linear Regression</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.266</td>\n","      <td>0.666</td>\n","      <td>0.968</td>\n","      <td>0.334</td>\n","      <td>0.032</td>\n","      <td>0.888</td>\n","      <td>0.885</td>\n","      <td>0.666</td>\n","      <td>0.882</td>\n","      <td>0.759</td>\n","    </tr>\n","    <tr>\n","      <th>7</th>\n","      <td>Support Vector Classifier (RBF)</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.280</td>\n","      <td>0.656</td>\n","      <td>0.977</td>\n","      <td>0.344</td>\n","      <td>0.023</td>\n","      <td>0.887</td>\n","      <td>0.898</td>\n","      <td>0.656</td>\n","      <td>0.916</td>\n","      <td>0.765</td>\n","    </tr>\n","    <tr>\n","      <th>8</th>\n","      <td>Decision Tree with Bagging</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.166</td>\n","      <td>0.750</td>\n","      <td>0.909</td>\n","      <td>0.250</td>\n","      <td>0.091</td>\n","      <td>0.882</td>\n","      <td>0.784</td>\n","      <td>0.750</td>\n","      <td>0.620</td>\n","      <td>0.678</td>\n","    </tr>\n","    <tr>\n","      <th>9</th>\n","      <td>Robust Regression</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.285</td>\n","      <td>0.631</td>\n","      <td>0.971</td>\n","      <td>0.369</td>\n","      <td>0.029</td>\n","      <td>0.874</td>\n","      <td>0.882</td>\n","      <td>0.631</td>\n","      <td>0.897</td>\n","      <td>0.740</td>\n","    </tr>\n","    <tr>\n","      <th>10</th>\n","      <td>Decision Tree with Boosting</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.262</td>\n","      <td>0.630</td>\n","      <td>0.952</td>\n","      <td>0.370</td>\n","      <td>0.048</td>\n","      <td>0.868</td>\n","      <td>0.851</td>\n","      <td>0.630</td>\n","      <td>0.824</td>\n","      <td>0.714</td>\n","    </tr>\n","    <tr>\n","      <th>11</th>\n","      <td>Nearest Neighbors</td>\n","      <td>Embedding Vectors</td>\n","      <td>0.322</td>\n","      <td>0.555</td>\n","      <td>0.968</td>\n","      <td>0.445</td>\n","      <td>0.032</td>\n","      <td>0.835</td>\n","      <td>0.857</td>\n","      <td>0.555</td>\n","      <td>0.893</td>\n","      <td>0.685</td>\n","    </tr>\n","    <tr>\n","      <th>11</th>\n","      <td>Naive Bayes</td>\n","      <td>Bag of Words</td>\n","      <td>0.344</td>\n","      <td>0.550</td>\n","      <td>0.983</td>\n","      <td>0.450</td>\n","      <td>0.017</td>\n","      <td>0.834</td>\n","      <td>0.876</td>\n","      <td>0.550</td>\n","      <td>0.945</td>\n","      <td>0.695</td>\n","    </tr>\n","    <tr>\n","      <th>12</th>\n","      <td>Nearest Neighbors</td>\n","      <td>Bag of Words</td>\n","      <td>0.348</td>\n","      <td>0.480</td>\n","      <td>0.948</td>\n","      <td>0.520</td>\n","      <td>0.052</td>\n","      <td>0.786</td>\n","      <td>0.803</td>\n","      <td>0.480</td>\n","      <td>0.832</td>\n","      <td>0.609</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["                               Name               Type  ...  Recall     F1\n","0                     Random Forest       Bag of Words  ...   0.858  0.897\n","0                           SciBERT        Transformer  ...   0.882  0.882\n","2                      PatentSBERTa        Transformer  ...   0.901  0.879\n","3                           BioBERT        Transformer  ...   0.901  0.879\n","1                            ALBERT        Transformer  ...   0.857  0.873\n","4                        DistilBERT        Transformer  ...   0.907  0.869\n","5                              BERT        Transformer  ...   0.894  0.865\n","6                             XLNet        Transformer  ...   0.901  0.866\n","1   Support Vector Classifier (RBF)       Bag of Words  ...   0.791  0.847\n","7                           RoBERTa        Transformer  ...   0.913  0.865\n","8                           ELECTRA        Transformer  ...   0.851  0.846\n","0   Doc2Vec USPTO Patent Embeddings                CNN  ...   0.853  0.848\n","9                          Big-Bird        Transformer  ...   0.901  0.848\n","1                   GLOVE (6B - 50)                CNN  ...   0.810  0.832\n","2               Logistic Regression       Bag of Words  ...   0.895  0.844\n","2                GLOVE (840B - 300)                CNN  ...   0.808  0.821\n","3      Gradient Boosting Classifier       Bag of Words  ...   0.872  0.824\n","4            Multi Layer Perceptron       Bag of Words  ...   0.812  0.810\n","5                  Ridge Regression       Bag of Words  ...   0.852  0.814\n","6       Decision Tree with Boosting       Bag of Words  ...   0.884  0.817\n","3                          FastText                CNN  ...   0.782  0.796\n","7                             LASSO       Bag of Words  ...   0.850  0.809\n","0            Multi Layer Perceptron  Embedding Vectors  ...   0.809  0.784\n","4                     No Embeddings                CNN  ...   0.749  0.763\n","1                     Random Forest  Embedding Vectors  ...   0.685  0.744\n","8        Decision Tree with Bagging       Bag of Words  ...   0.793  0.769\n","2      Gradient Boosting Classifier  Embedding Vectors  ...   0.837  0.762\n","3               Logistic Regression  Embedding Vectors  ...   0.906  0.771\n","5                             LASSO  Embedding Vectors  ...   0.892  0.764\n","4                  Ridge Regression  Embedding Vectors  ...   0.895  0.765\n","9                 Robust Regression       Bag of Words  ...   0.827  0.751\n","10                Linear Regression       Bag of Words  ...   0.800  0.743\n","6                 Linear Regression  Embedding Vectors  ...   0.882  0.759\n","7   Support Vector Classifier (RBF)  Embedding Vectors  ...   0.916  0.765\n","8        Decision Tree with Bagging  Embedding Vectors  ...   0.620  0.678\n","9                 Robust Regression  Embedding Vectors  ...   0.897  0.740\n","10      Decision Tree with Boosting  Embedding Vectors  ...   0.824  0.714\n","11                Nearest Neighbors  Embedding Vectors  ...   0.893  0.685\n","11                      Naive Bayes       Bag of Words  ...   0.945  0.695\n","12                Nearest Neighbors       Bag of Words  ...   0.832  0.609\n","\n","[40 rows x 12 columns]"]},"metadata":{"tags":[]},"execution_count":5}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":114},"id":"NwzypV6T_hO5","executionInfo":{"status":"ok","timestamp":1627938900323,"user_tz":420,"elapsed":6,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"7f9b4a41-8fa6-4496-a0fb-67b219ea1cff"},"source":["# Best Performing Model #\n","\n","Combined.drop(\"Unnamed: 0\", axis = 1).sort_values(\"Accuracy\", ascending = False).head(1)"],"execution_count":null,"outputs":[{"output_type":"execute_result","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>Name</th>\n","      <th>Type</th>\n","      <th>Share</th>\n","      <th>True-Positives</th>\n","      <th>False-Negatives</th>\n","      <th>False-Positives</th>\n","      <th>True-Negatives</th>\n","      <th>Accuracy</th>\n","      <th>AUC</th>\n","      <th>Precision</th>\n","      <th>Recall</th>\n","      <th>F1</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>Random Forest</td>\n","      <td>Bag of Words</td>\n","      <td>0.183</td>\n","      <td>0.94</td>\n","      <td>0.965</td>\n","      <td>0.06</td>\n","      <td>0.035</td>\n","      <td>0.96</td>\n","      <td>0.922</td>\n","      <td>0.94</td>\n","      <td>0.858</td>\n","      <td>0.897</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["            Name          Type  Share  ...  Precision  Recall     F1\n","0  Random Forest  Bag of Words  0.183  ...       0.94   0.858  0.897\n","\n","[1 rows x 12 columns]"]},"metadata":{"tags":[]},"execution_count":6}]},{"cell_type":"code","metadata":{"id":"bqYh_lXK5Qgd"},"source":["Combined.to_csv(\"./Output/All Model Performance Combined.csv\")"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"TFMCrqi7BeyM"},"source":["The Random Forest Model performs best on two important metrics: \n","\n","- **Accuracy**  [ *Total Share of Observations Correctly Classified* ] \n","- **False Positives** [ *Total Share of Postive Values (AI Patents) Incorrectly Classified* ] \n","\n","In our case, we are particulary concerned about False Positives. We would prefer to have a smaller share of False Positives but a higher share of True Negatives than the opposite (this would mean our subsequent results are a lower bound and more conservative than the alternative).\n","\n","For other applications, we might want to focus on the overall accuracy or recall (which reflects both False Positives and True Negatives)\n","\n","*Note.* An additional benefit of using a Bag-of-Words (BOW) model is that it is fast (computationally cheap) to classify the entire database of AI patents. In comparison, Transformer models are typically much slower in classifying observations. Researchs may want to consider this difference if they use datasets with a vary large number of observations. "]},{"cell_type":"markdown","metadata":{"id":"MsgeTnz1Gufr"},"source":["# X.3. Load Classification Results [Random Forest Model Only] ##\n","---------------- \n"]},{"cell_type":"code","metadata":{"id":"HpNRcOMDGpH8"},"source":["# Classified Training Data #\n","\n","CTData = pd.read_csv(\"./Output/Classification Output/BOW Classification Results.csv\")\n","\n","CTData = CTData[['id', 'Actual', 'Random Forest']]\n","\n","CTData.columns = ['id', 'Actual', 'Predicted']\n"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"ewiZUFxAv9_v"},"source":["### X.3.1 Plot Confusion Matrix \n","---- \n","\n","Below, we plot a normalized confusion matrix. This provides an intuitive representation of the share of correct and incorrect predictions (e.g. False Positives).  "]},{"cell_type":"code","metadata":{"id":"y2x4rwvkG6Vn","colab":{"base_uri":"https://localhost:8080/","height":422},"executionInfo":{"status":"ok","timestamp":1626218773617,"user_tz":420,"elapsed":216,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"fb16d552-99f6-4c8b-b210-89bc3238a5f8"},"source":["# Plot confusion Matrix # \n","\n","from sklearn.metrics import ConfusionMatrixDisplay\n","from sklearn.metrics import confusion_matrix\n","\n","plt.rcParams['font.size'] = '16'\n","plt.rcParams['font.family'] = 'serif'\n","\n","cm = confusion_matrix(CTData['Actual'].values.tolist(), CTData['Predicted'].values.tolist())\n","cm = np.round(cm/cm.sum(axis = 0), 2)\n","\n","disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=['Not AI', 'AI'])\n","fig, ax = plt.subplots(figsize=(7,7))\n","disp.plot(cmap=plt.cm.Blues, ax = ax)\n","plt.show()"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 504x504 with 2 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"markdown","metadata":{"id":"avwCVZNvGonu"},"source":["The confusion matrix reflects the share of values that were classified either correctly or incorrectly. This confusion matrix is normalized by the number of predicted labels. Therefore, we can see that of those where **Predicted Label** is **AI**, 92% were correctly classified, and only 8% were misclassified. \n","\n","\n","For more in-depth comparisons, we need to also observe the probability of being AI related, rather than just the binary classification."]},{"cell_type":"markdown","metadata":{"id":"8uwn4BJKwd07"},"source":["### X.3.2 Plot Receiver Operating Characteristic (ROC) Curve\n","\n","--- \n","\n","ROC Curves provide a representation of model accuracy that shows how the model is able to correctly identify positive values (i.e. AI patent). There is an inherent tradeoff in that if we classify more observations as being AI patents, we may capture more real AI patents (i.e. True Positive Rate), but also may incorrectly classify non-AI patents as being AI related (i.e. False Positive Rate). \n","\n","A model that performs well does a good job of balancing this tradeoff and can achieve a high rate of True Positives, with a low rate of false positives (this can be seen as a steep vertical line followed by a flat line). Any point on this curve can be interpreted as the level of false positives one would expect to get for a given level of true positives (e.g. 80% True Positive Rate coresponds to 3% False Positive Rate). \n","\n","In order to compare different models we can compute the area under the curve (AUC). This indicates the share of the geometric area covered by the curve. A value of 1 means a perfect classification (100% True Positive Rate, with 0% False Positive Rate). \n","\n","Below we plot the ROC Curve for the Random Forest [BOW] Model and report the AOC. AOC values for all models are reported in results summary tables. "]},{"cell_type":"code","metadata":{"id":"u4t_AKEBLy0W"},"source":["CTDataP = pd.read_csv(\"./Output/Classification Output/Random Forest - BOW [Predicted Probabilities].csv\")\n","\n","CTDataP = CTDataP[['id', 'Actual', 'Random Forest']]\n","\n","CTDataP.columns = ['id', 'Actual', 'p1']\n"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"ROUWBp9nQ7ez"},"source":["from sklearn.metrics import roc_curve, auc\n","\n","fpr = dict()\n","tpr = dict()\n","roc_auc = dict()\n","\n","fpr, tpr, thresholds = roc_curve(CTDataP['Actual'].ravel(), CTDataP['p1'].ravel())\n","roc_auc = auc(fpr, tpr)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"CglHFo4kQ7br","colab":{"base_uri":"https://localhost:8080/","height":429},"executionInfo":{"status":"ok","timestamp":1626218774300,"user_tz":420,"elapsed":253,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"d8ff6a96-0c28-482e-ebcc-7eec013ac58a"},"source":["plt.figure(figsize= (10,6.5))\n","plt.plot( fpr, tpr, color='dodgerblue', lw=3,\n","          label='ROC curve (area = %0.3f)' % roc_auc)\n","plt.plot([0, 1], [0, 1], lw=3, linestyle='--', color = 'black')\n","plt.xlim([0.0, 1.0])\n","plt.ylim([0.0, 1.05])\n","plt.xlabel('False Positive Rate', size = '15')\n","plt.ylabel('True Positive Rate', size = '15')\n","plt.legend(loc=\"lower right\", fontsize = 'large')\n","#plt.title('Receiver operating characteristic example')\n","plt.ylim(0,1.01)\n","plt.xlim(-0.01,1)\n","\n","plt.show()"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 720x468 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"markdown","metadata":{"id":"N60-Bi_ryFEq"},"source":["### X.3.3 Precision Recall Curve \n","\n","--- \n","\n","One misleading aspect about ROC curves is that they may be overly positive (optimistic) when there is class imbalance. This is because the model can naturally achieve a high True Positive Rate by favoring the dominant class. \n","\n","A precision recall curve plots the precision (share of actual positives correctly classified among all positive values) with respect to the recall (share of positives predicted positives, correctly classified). The mean of these two values is the F1 Score. \n","\n","A well performing model has high precision and high recall. Whereas a low performing model (typically with imbalanced classes) has a high precision, but low recall. This coresponds to a graph that is flat and then declines sharply as it approaches high recall values. Reading values on the graph can indicate at what level of precision, we can expect to achieve which level of recall (e.g. 90% precision, 90% recall). \n","\n","Below we plot the Precision-Recall Curve for the Random Forest [BOW] Model and report the Average precision Score. Like the AUC metric, it reflects how well this model performs and allows us to comapare different models. We illustrate f1 thresholds using grey lines on the model. This indicates the domains where we would achieve a high f1 score which may be a more intuitive metric. "]},{"cell_type":"code","metadata":{"id":"EI6c7s5JQ7X9"},"source":["from sklearn.metrics import precision_recall_curve\n","from sklearn.metrics import average_precision_score\n","\n","precision, recall, thresholds = precision_recall_curve(CTDataP['Actual'].ravel(), CTDataP['p1'].ravel())\n","average_precision = average_precision_score(CTDataP['Actual'].ravel(), CTDataP['p1'].ravel(), average=\"micro\")"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"ydk9QpNnRVE4","colab":{"base_uri":"https://localhost:8080/","height":426},"executionInfo":{"status":"ok","timestamp":1626218774553,"user_tz":420,"elapsed":266,"user":{"displayName":"Annoymous Researcher","photoUrl":"","userId":"13222073037946805330"}},"outputId":"ff426f84-05c0-4631-f592-4b81df829e17"},"source":["eplt.figure(figsize= (10,6.5))\n","plt.step(recall, precision, where='post', lw = 3, color = 'dodgerblue',\n","         label='Average Precision Score AP = %0.3f' % average_precision)\n","f_scores = np.linspace(0.2, 0.8, num=4)\n","lines = []\n","labels = []\n","for f_score in f_scores:\n","    x = np.linspace(0.01, 1)\n","    y = f_score * x / (2 * x - f_score)\n","    l, = plt.plot(x[y >= 0], y[y >= 0], color='gray', alpha=0.2)\n","    plt.annotate('f1={0:0.1f}'.format(f_score), xy=(0.82, y[45] + 0.02))\n","plt.xlabel('Recall', size = '15')\n","plt.ylabel('Precision', size = '15')\n","plt.ylim([0.0, 1.05])\n","plt.xlim([0.0, 1.0])\n","plt.legend(loc=\"lower left\", fontsize = 'large')\n","plt.show()"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAnsAAAGZCAYAAADxUPeDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeZxjZZno8d97TpKTPVWpVDWrLEqzyrQKKIIiqFdRaR0QRR115DoO6ihXRscZdbjMKIKjXnXU0VFn3B1QQEFHQVR0ABVpBBtBkF2Upqqy76ks7/3j1Ekn1dW1Zs/z/XzyqeQk5+Sp6k7y5Dnv+7xKa40QQgghhBhNRr8DEEIIIYQQ3SPJnhBCCCHECJNkTwghhBBihEmyJ4QQQggxwiTZE0IIIYQYYa5+B9ArsVhMH3zwwf0OQwwwrTW1Wg2Xy4VSqt/hjB35+4+nRqNBvV7HNE0MQ+oPQjhuu+22uNZ6uhPHGptk7+CDD2bHjh39DkMMsGq1yvz8PJOTk/h8vn6HM3YajQazs7MEAgHC4XC/wxE9lEgkqFarTE9PY5pmv8MRYiAopR7p1LHka5QQi5yqgvSe7A/DMHC73VQqlX6HInosEomgtSabzfY7FCFGkiR7Qixykr1Go9HnSMaX1+ulWq1Sr9f7HYroIZfLRTAYpFQqSbIvRBdIsifEIqUUSilJ9vrIsiwA+cAfQ8FgEJfLRTqdluq6EB0myZ4QLQzDkGSvj9xuN6ZpSrI3hpRSTExMUK/XyeVy/Q5HiJEiyZ4QLSTZ6z/LsqhUKlLdGUMejwe/30+hUKBarfY7HCFGhiR7QrRQSkmS0WeWZdFoNOTDfkyFw2GUUmQyGXktCtEhkuwJ0UIqe/0n4/bGm2EYRCIRFhYWKBaL/Q5HiJEgyZ4QLSTZ6z/DMPB4PJLsjTGfz4dlWWSzWWq1Wr/DEWLoSbInRAtJ9gaDZVksLCzIv8UYm5iYQClFKpWS07lCbJIke0K0MAwDrbV8uPSZnMoVpmkyMTFBtVqVZstCbFLPkz2l1L5KqWuVUvJpKgaOsyarVJT6y+12YxiGJHtjzuv1EgwGKRQKlEqlfocjxNDqabKnlDoT+AXwxA3u/3+UUncrpXYqpX6tlHpZZyMU405W0RgMSqlmCxYx3kKhEB6Ph0wmIyurCLFBva7svRt4PnDzendUSv098D7gDK31sYvH+pZS6vTOhjga6vU6u3btktls6yTr4w4Oy7Ko1+vSgmXMOc2WARm/J8QGuXr8fCdprWvOqbK1UkpNAP8IfFRr/QCA1vp6pdQPgY8AP1jtGMUq3P74BiLugSkfPCHS2WM6/eKkQrU+UtkbHK3j9txud5+jEf3kcrmIRCKkUilyuRzhcLjfIQkxVHqa7GmtNzqH/oWAH7hhyfafAB9RSh2htb5npQM8kIKXXb7BZ++Bvz0R3n5C547nJNTyLXh9RjHZ0xoaGjSLPzXUNdQa9jYW72t9nHZ+ttzv7Otsd/5r6db7gYUamEbLY5fss/Q5l963+zEmuZQbnSoTiAT3eExuAfzuZfZrPX5rnEv2d7axZN/l4m47ZstjsxUwFHiXvJMud7w9rrfEAXvG2Rbf0m17eWzb9SXHBfC5YftW2CfI0PH5fFQqFfL5PJZlNb8MCCFW1+vK3kYdu/jzoSXbH2q5f49kTyn1JuBNAJ4Dn9a14DrhO/d0PtkbltUgnOSj3oByHYoL9u3G4rbW64kSuIzFbQ1oYG9vaJgrgOXanbTU9e7rlTo8nIZ9g7uP11jmcbWGopKCe/INnrilPflp3cdJbm75IxweA8ts3770uM39NNwxCxELpgOLH/5LjtlMKLT9+7UeI1GCoAcU7QlYY/GfeWlSN/j/+ivzNyx8jQJJs4FW0jygEy6+EbZtWTkRbyyzXWuoNuCMrXDGYXtP4LdO7Zn8dkokEqFarZJKpZiensY0ze48kRAjZliSvdjiz6WrYzvz8aeW20lr/TngcwAThx6nt23pTnAbVanD7+L29VoXCklr6RlXWLArJE4ykalAuWYnULvykCiCe/H9dLYAmTKELPv+amP343bl4MCI/XtU67Bjl/2GH/Pbj6ktPv4PGfC57A+F1kRucBhM1RQlo8Gt8bXtEX90/c+SqdiXjcgvbGy/YbSgvPjI49YLLChvv8MZGXfMbnzfT99qX1byrZfDCftv/Dn2xhm/F4/HSafTTE0t+9YvhFhiWJK9TXtSFK4+p99RtHs4Dad8eWP71htwd9yuboFdEXskbSdUu/KQq8DcnKLY0DxUhgPDsFCHh9IwX4QDQvDHpanzJu2ca7+drdjVtqVKA94QXyuFGvqaWDuFfbpRKfvnwuKkxtBilZDF7Qr7MYrd12nZt3W7s1/z2Ivb8wv2/7HDoivvs9x9zvM1b2s3vgWDhlGmZnnb9wd++Sc4bl/79CRLYl72uEuus+QxS7ft7XrrttmC/bo7KLL8cZc+L+w9nrbrrf94a3xs63OyJO5P/oqeOfsK+Pqf29edKrdTia/W7Qp1yLO7cu1U550vgY3Fqna5Zn+RbK/Iu6lXItRLabQnD55g8/H1hn28sAWv+zOI+nr3OwsxyIYl2XNqLCEg0bLdGaWbYIQ8krbfDG9/3P4QvW0XzBfscYfzRZjwQrq8+nEidQNNg6wJj2Ta7+t0ordZhrJPzyrsiue03z4layr7Yhj2z1wFHsvDMw7YfZ9yHqPgnoSdZMR8i9sNe8q588F4fxKetq99PAP7fqXarzfyBobRIKng0An7uM5FqSW3sSujUd/uGFofay63H1DTi4lWy3Haki21O+6lSZhptP9eS+9vfb6lScbwUaRSFgsLFbYMWGV+2PztM+D798OWwMrJu7HMdqXgJw/DN+9afJ0ss9/d8+3P95pvd/O38ROqV/DoHBnTQ0159njEx26xh1dcdhY8ZZ9hfx0IsTnDkuztXPx5MPBwy/ZDltw/tB7JwBGfXlvVay2JHoBm7RWqfYP2mzzAn3Lw1H3BY8Afs7B/2E6gwK7WuQx7XI5pgHsxCQO7urFv0D7t6zLsb+Ex32LSZoBrMfnyuyHgtp/PNHYnbIMiHjdQqoGcIRoMlmVRKpWoVqsyK3cTlIIXH7bx/bdOwXkrDH2uN+DQT278+OuVNyJEGlVCjRRpY3rZMZ2VOvz5N+3rh07YX7IaDft9bCYI73i6PX72mQfa70VCjKqBTPaUUlNATmvtjE66FigCzwF+2vLQU4G7V5uJOyw2enrz2Bn7Z7xoJ1xPnrFbzRweMDB1lYP3t6tI+4ft5Mxt2glXyGP/DO75pXisGYYhzVsHiLRgGQ6mYZ+6/ffb7NOvTiXbNNor3qayk7D7k3D8fu33LXdpaPusxrEzi9sWq9qGMqA+AcUEpjuNOxhFKfjgTcvH92C6/XY2CW9tadq1dWr3aeCGtocjhCy44Bm7J5DVFscZHz4Fx+3XtT+lEB03cMmeUuoQ4HfYbVZOB9Bap5VS7wf+Vin1Fa31g0qp5wEvALb3L9rNOTAM+4fsStpynjgJqTKcdaRdzTswDDMB+41m/5D9RuRf4bMvnVZUKlpOf62TYRjSyHeAmKaJ2+2mUqkQDA5hz5AxcvIT7EvveMjnQ2SzWSKRAoFAgL9+Gtz6J3j5Fes70u+XGQwUL8Hbrt37PlHf7jGGuQV7+MlfPdW+rYBnHwRHTa8vDiG6oafJnlLqw9graDxh8fYdi3ed0FLFKwFJ4LHWfbXWlyqlysD3lFI1oA6crbVetaHyoDIN+O9X2WNhFurwjP1hS9D+5mt14F9mLbNxxZ6GpWXNOLEsi0KhgNaa9TZlF6MtGAyysLBANpvF7Xbj8Xg4fn945Hz7S/Jv5+wvyU5V8ZEMfPNuO0m79oHNPXdyyXK988X2yuLHb4Fb32h/MRein9S4fKgdd9xxeseOHf0Oo6dyuRy5XI59991XPiDXQf5ug6dSqZBIJIhGo3i90oJFtGs0GsTjcRqNBrFYDJdrbd+Waw14MGUnfq0Tmx5Kw9d22u2jXIsTolwG3Dm3fAVwLZ4Uhc+8yD5dLMRaKKVu01of14ljDdxpXNE5reu8StKydvJ3GzwejwelFJVKRZI9sQfDMIhGo8TjcZLJJLFYrPk6Xokz2WypJ0Xh+Ycuv0+5ZlcMW8cg3p+C/75v92z4L9y+5373J+H5X9t9+7yn2V0FTjlo9+Q4IbpFkr0RJkumbUzrkmlr+cAQ3aeUwrIsyuUy4XBYknCxB5fLxeTkJMlkklQqRTQa7cr/E69rz+XmjvfZk00ct/zJrgKu5LO32ZfPvwT+1xM7HqYQbeSTbISN4jqvvSB/t8Hk9/up1+tUKhtcekSMPMuyiEQiVCoVMpnM6jt0yfdeBb84F371v+Hys1Z+7F99D57zZbjj8d7EJsaTVPZGmFT2Nsb5u0myN1gsy8IwDEqlkpzKFXvl9/up1Wrk83lcLlffZnDvF7J/bgnak0XyC/YYwR/cD//0s/ZWWw+l4aWX29dfebQ9m/jI6d3LT1Yb8GgGXrwVDpnYvY74HzJ2L9NqY/djtwTtU9DVut3NoVi128XUGruXqFTYrWNi/p7/WUSfSLI3wiRp2RhncXXptTdYlFL4/X4KhYKcYhcrCofD1Ot1stksLpdrIL4cOP1MX3UMnHYIvPbbcO8ykz0uv8v+ubQvIKy+JrHjop+t7XFHTcOzFlvlBNzw50fAEyJr21cMF0n2RljrRAOxdqZpYhgGtdqAL+I7hnw+H/l8nlKpRCAQ6Hc4YoBNTExQr9dJpVLEYrGBasi9JQDXvcae2HHpTfCjh/oTx93z7cvc/b9fwvat8LIj4LmH7H0/MXwk2RthUtnbOJfLJcneAHL6qBWLRUn2xIqUUkxOTrbN0HWq9oNAKXsZyv/Ybk/meMPVdkPqp+xjr3SUrcBBE/b1r+y0T8MWq/Y66SceaM8Cbmh4NGuv9/1oFq6+1+4p6F5sFeMy7NVHjorZ213G6onlNb+3L6c/Cf7leRCWHoEjQfrsjTCtNbt27SIcDsvKA+uUTqepVCpskeVHBk6hUCCTyTA9PT1Q1RoxmKrVKolEAtM0icViMpMbu5p3w8O71zW/5Oa9P/avn2a3m3nl0XC0rAbSU9JnT6yJUgqllFT2NsDlclEsFmVs2ADy+Xxks1mKxSKRiAwwEitzu91tLVkmJyfHPuE7arp9GbdXHQM3PQrv/pG97Furf7/N/vnl39g/X/BE+LtnwsETdrWxocEy7cbTYnBJsjfiZOmvjXE68NdqNTweT5+jEa0Mw8Dr9VIqlaTnnlgTy7IIh8NkMhlyuRzhcLjfIQ2UiBdefJh96vZHD9rtYPbmugfsy1LvOtGeBbx1yj5FLQaLJHsjTtbH3RhJ9gab3++nVCpRLpfx+Xz9DkcMgUAg0GzJYpqmjPlchqHsBs+3vwm+utO+/W877LGCq/nwL+yLcxywq35uwx6L+Iz94TVPlnWC+0WSvREnlb2NMU0TpZRM0hhQlmVhmibFYlGSPbFmrS1ZTNMciJYsgyjqg/Ofbl9/2wmwKwd3x+Hca9a2f6PlI6fasMcH3vCwPTbw3c+Etxzf6YjFaiTZG3FS2dsYpRSmaUqyN8D8fj+5XI56vT5QsyzF4HJm6CYSieb4PUn4VrdvyL48cr69LrBSELFgoQ6X3gw7HoPfzK7tWB/6Ody2C159DDx3L+sPi86TZG/EyQSNjXO73VSrazh/IfrCSfaKxSKhUKjf4YghoZQiGo02E75oNIplybnFtZpoyY09Jlz4bPu61vBYzm7vEvPD43n45Z/gpj/AVfe0H+NHD8HNj8Ktb5TTur0i82dGnJzG3TiXy0W9Xpe/34AyTRPLsiiVSv0ORQwZwzCYmprC5XKRTCZlveUOUAr2D9vLtZmGff2sI+FjL7Argi94YvvjSzU481vwjTv7E++4kWRvxMlp3I1zuVxorWXZtAHmrIMqH9ZivZyEzzRNkskkCwsLq+8kNuyzL4bLz2rf9vsE/MNP4KBPwDuug/99DVxzL+Tk5dxxkuyNOKnsbVzrjFwxmLxeL4ZhUCwW+x2KGEKS8PWOoeAZB8CZRyx//1X32Kd333YtHPNZeMUVEJeXdcdIsjfiDMNAay0J3wY4yZ6M2xtcSil8Ph/lclkq2GJDTNNkamoKwzBIJpPyeu+yS54LXzgDXvjElR93y5/gaZ+3f4rNk2RvxMn6uBsnM3KHg9/vR2stY/fEhjkJn1KKRCIhCV8XeV128+V/fwl8+aX26htvfprdjHk5r7gCClJw3TRJ9kacs9SXVPY2xuVySbI34NxuN263W5I9sSmS8PXecw6G3/8N/P3JcP1f2BM5vvXyPR932V09D23kSLI34qSytzmS7A0Hv9/PwsKCfECLTXG5XG0Jn7z2e++E/eF3b2nfdtuu/sQySiTZG3FS2dscmZE7HHw+H0opqe6JTZOEr//8bviHk3ff/u/74KKf2Q2dxcZIsjfipLK3OTIjdzgYhoFlWRSLRfliIzbN5XIRjUbRWkvC1ydHxdpvf/EOeO23+xPLKJBkb8RJZW9zJNkbHn6/n0ajIT33REe43W6mpqaaCZ8MEeitk58AB0fat+2cg/+4He5L2it2iLWTZG/EOZU9SfY2xjRNDMOQZG8IWJaFaZrSc090jJPwASQSCfki0UOGgp++Hj75wvbt//w/8LyvwsH/Cq//DlRlhM2aSLI34uQ07ubJJI3h4PTcq1QqMsZSdIzb7SYWizX78Mm40N5RCrYfDi87fPn7f/oIPOlT8LTP2dU+sXeS7I04pZSsorFJkuwND+m5J7rBNE1isRhut5tUKkWhUOh3SGPl0ufaa+vGfMvfHy/Z1b689OPbK0n2xoCsj7s5LpeLer0uf8Mh4HK58Hg8cipXdJyztJrX6yWTyZDL5fod0tjwueFzL4Hb3mT34rv0ucs/7ocP9jauYSLJ3hiQyt7myCSN4eL3+6nVarLOqeg4pRSTk5P4/X5yuRzpdFreW/vgVcfYSd/db27f/o7r4J54f2IadJLsjQGp7G2OJHvDxem5J9U90Q1KKSYmJgiFQhSLRVKplCR8fRLw7LntBV+HeyXh24Mke2NAKnubY5omSilJ9oaEM1GjVCrJlxzRNaFQiEgkQrlcJpFIyP+1PrnxL/fcdtOjvY5i8EmyNwaksrc5SimZpDFkAoEAWmsZSC+6KhAIMDk5SbVaJR6PyyzwPnhCBH76uvZt//w/0odvKUn2xoBU9jZPkr3h4na78Xq9FAoF+aIjusrn8xGNRmk0GsTjcWm+3AeHTELYWrLtX+H790FF3rYBSfbGglT2Ns+ZkStJ8/AIhUI0Gg2p7omusyyrrflyuSyLuPbad89pv62BN38f/u5HfQln4EiyNwacyp4kKhvncrnQWstpmiEi1T3RS07zZdM0SSaT0pqlxw6egMvP2nP7d+6FR9K9j2fQSLI3BmR93M1zZuTKKZrh4lT3ZGau6AWn+bLP5yOXy5FMJuWLRg894wA74XvuIe3bn/81SI55n3VJ9saArI+7edJ+ZTg51b18Pi8fuqInnF584XCYSqVCPB6X940eesYB8JkXwVTLahuVOvzv7473ChuS7I0BWR9385RSmKYpb9pDKBgMSnVP9FwwGGybuCHj+HrHcsFXXta+7de74OjPQGpMK3yS7I0BOY3bGTIjdzh5PJ5mdU9eA6KXLMtienpaxvH1wTEz8Nbj99z+s0d6H8sgkGRvDEhlrzMk2RteTnVPZuaKXpNxfP3ztuP3XEe3Pqbf9yTZGwNS2esMmZE7vDweD5ZlSXVP9IWM4+sPn9teR/fMI/odSf9JsjcGpLLXGTJJY7hJ3z3RbzKOr/92zvY7gv6QZG8MSGWvMyTZG25S3RODYOk4vmw2K/8fe+hLvxnPvnuS7I0BpRRKKansbZJpmhiGIcneEJPqnhgEzji+QCBAPp+X07pdduIB7bc/eFN/4ugnSfbGhKyP2xkySWO4SXVPDAqlFJFIhGg0Sr1eZ35+XtoDdcnZR7XfvvYBuxXLOJFkb0zI+ridIcne8JPqnhgkXq+X6elpPB4P6XRaZut2gVJ7LqX259+EC34I4/KdT5K9MSGVvc5wuVzU63V5Mx5iUt0Tg8Y0Taamppqzdefn56lUKv0Oa6QcPQ0Bd/u2K38H1z/Yn3h6TZK9MSHJXmfIJI3RINU9MYiCwSCxWAylFIlEQiZvdFDIguv/Ao7bt337Tx7uSzg9J8nemJDTuJ0hyd5ocKp7hUJBPkzFQHG73UxPT8vkjS7YPwxXvgK2b9297b9+C+Ux+PNKsjcmpLLXGaZpopSSN98REAqFqNfrMiheDByZvNFdLzqs/fadY9B7T5K9MSGVvc5QSskkjREhY/fEoFtu8oas4LN5zzuk/XZtDF7+kuyNCansdY4ke6MjGAxKdU8MNGfyRiQSoVKpMDc3J8MPNsltwvH77b59zpXwx+xon86VZG9MGIaB1lreIDrAmZErf8vhZ1mWVPfEUAgEAszMzODxeMhkMiQSCarVar/DGlpTvvbbJ30RDv80XHVPf+LpNkn2xoSsj9s5LpcLrbVU90aEU93L5/P9DkWIFTlVvsnJSWq1GvF4nFwuJ19UNuD/nrL89ndcB5/7dW9j6YWeJntKqRml1NeVUvcuXq5QSh2w+p6glNpXKfUFpdQ9SqmdSqm7lFLvUUq5V99byPq4nSMzckeLZVn4fD7y+bz8m4qh4PP5mJmZwev1ksvlmJ+fZ2Fhod9hDZX9QnYrluVcfCMUR6xo2rNkTynlAa4HPMDRwFFAAbhBKRVcZV8D+D7wDOAkrfWxwF8A/whc0s24R4VU9jpHkr3REw6HUUqRyWT6HYoQa2IYBpOTk0SjUbTWxONxMpmMvMevw9YpeOR8uPete9731Z29j6ebelnZez1wLPBurXVNa10H3g0cCrx5lX2PALYB/6G1TgBorW8Hfgi8pnshjw6p7HWOUgrTNCXZGyGmaRIKhahUKpRKpX6HI8SaOTN2A4EAhUKB+fl5yuVyv8MaKl6XnfS1+uBNUBqh6l4vk72zgD9orZuLk2itHwfuXrxvJc6nqmvJdhdgdizCEeZU9iTZ6wyZkTt6AoFAc/C7VEfEMDEMg0gk0lx9I5lMkkqlpE3LOp1/QvvtI/4NGiPykdnLZO9Y4KFltj8EPHmlHbXWvwe+Afy1UupgAKXUacDzgU92NMoR5VT25EOsMyTZG02RSAStNdlstt+hCLFuHo+H6elpQqEQ5XKZubk5mcCxDm9/+p7bfp/ofRzd0MtkLwbkltmeBfxKKd8y97V6Pfa4vfuUUo8B3wH+j9b6/XvbQSn1JqXUDqXUjvn5+Y3GPRKkstdZzoxc+eY8WtxuN4FAgGKxKAvRi6GklCIUCrVN4Jibm5PhCWvgMmDHG9u3VUekPjIUrVeUUl7gBuAE4GCt9X7Ac4B/UEq9d2/7aa0/p7U+Tmt93PT0dG+CHVAyQaOzZJLG6AqFQrhcLjKZjHw5EkPLNE0mJyeJxWIYhkEqlSIej0tvvlVMB+DolnThpZfBI+n+xdMpvUz24kBome1hoKi1Xulrx7nAycC7tNZ/AtBa/xr4CPB+pdS2Tgc7apRSsopGB0myN7qcdUlrtZr03hNDz+PxEIvFmJiYoFarMT8/Tzqdli/+K/C3NHSra7jsrv7F0im9TPZ2Agcvs/0Q4M5V9nXG9N23ZPvvAQUcv6nIxoSsj9s5pmliGIYkeyNKeu+JUaKUwu/3MzMzQyAQoFQqMTc3JyvH7MW5S8pH2REY0dHLZO8q4CBnggWAUmoLcCRwZesDlVJbFnvrOeYWfz5hyTEPWvw5IkMou0sqe50lkzRGm/TeE6PGmbU7PT2N2+0mm81Kq5ZlvOgwuGgvK2wMq14me1/CruB9SCnlWkzmLsWejfsZ50FKqZOAx4BPL9k3B3xAKRVafNwTgHcCDwI/6EH8Q08qe50lyd5oM02TcDhMpVKhWCz2OxwhOsblcjE1NUU0GgUgmUySSCRkFY4Wpup3BJ3Vs2RPa72A3Sqljt1b73fY4/VO01q3DozJAxlgV8u+D2FPzkgCtyqldgLXLV5OWmW8n1gklb3Ocrlc1Ot1SaBHmN/vx+PxkM1m5d9ZjBynIXMkEqFarRKPx0kmkzKJYwQtbVLcVVrrWeDVqzzmN0B0me33AOd0KbSxYBiGvIg7qHWShsfj6XM0olsmJiaYn58nm80yMTHR73CE6CilFIFAAL/fT6FQIJ/PMz8/j8/na85MF8NvKFqviM6Qyl5nyYzc8eByuQgGg9J7T4w0pRTBYJCZmRmCwSDlcpn5+XkymczY9xOtjUBRX5K9MSJj9jrLNE2UUpLsjYFgMCi998RYMAyDcDjMzMwMfr+fYrHI3NzcWA9luOwuKA/527wke2PEqezJh1VnKKVkksaYkN57YtyYptmcueu0IXKWXxuHpG+fYPvtHY/1J45OkWRvjMiSaZ0nyd74sCwLv99PPp+Xsa9ipD366KO85CUv4cgjj+Soo47i3HPPZXp6mmQyyfbt2zFNc+STvlMPbr897KdyJdkbI4Zh/3NLstc5TrInf9Px4PTeS6fT8m8uRtZb3vIWcrkcd911F7fffjt33nkn3/3ud3nRi17EH//4RwByuRyzs7Nks9mOjOm77bbbOOWUUzjmmGM4/PDDeec737nm/n87duzg9NNP58gjj+TJT34yJ5xwAt/61rc2FY/bhFMOWv1xw0KSvTEi6+N2nkzSGC+GYTAxMUG1WpVmy2Jk3XjjjZx44okYhkEgEGDnzp186EMf4vrrr+ekk04CYGZmBq/XS6FQYG5ublMTOe677z5OPfVUzjzzTH77299yyy23cN111/GGN7xh1X0ffvhhnvvc5xKLxbjzzju58847Offcc3nFK17Bd7/73Q3FM4ok2RsjUtnrPEn2xo/X6yUUClEsFqXZshhJmUwGr9fbvO3z+bj55ps57LDDmttcLheTk5PNMX3ORI50Or3u98OLLrqIaDTK29/+dsBud3ThhRdy2WWXceutt6647/e//32y2SwXXHBB8/34vPPOIxwO841vfGNdcYwySfbGiFT2Ok+SvfEUDAaxLItMJiPj9xRSg2sAACAASURBVMTIuOaaa9i2zV4Y9rOf/Szbtm3jda97HcBe++25XC4mJiaas3eddXdTqdSaXhu1Wo2rr76aU045pfkZBXDaaacBcOWVV+5t17a4Wt+DtdY0Go2xbxnTSroljhGp7HWeMyNXPvDHi1KKyclJ5ufnSaVSxGKx5utLiGG1fft2tm/fjlKK8847j4suumjN+zqzd0855RQajQaNRgOtNYZhYBhGWyIHcMcddwDw4IMPUigUOOSQQ9run5qaIhQKsXPnzhWf95xzzuFjH/sYH/jAB/j617+O3+/nkksuoVKpcN555605/lEnyd4Ykcped1iWRbFYRGu9xxuaGF2GYTA5OUkikSCdTjfXGRVinDlJXKPRoFgsUigUqNfruFyu5kodre+T8XgcgFAotMexwuEwiURixecLh8P8+Mc/5g1veAOxWIxgMEgkEuH666/nlFNO6djv9etdMB2Ao2IwjG/z8lV0jEhlrzssy0JrLYuIjyGPx0M4HKZcLkv/PSFaGIbRXJFjcnISpRSZTKajM3gB7r33Xk444QQOOuggkskkc3NzXHzxxZx55pn84Ac/6MhzAHziV/Cib8Crr+rYIXtKkr0xopRCKSWVvQ7zeDwopWQprTEVCATw+Xxks1n5PyDEEkopfD4f09PTxGIxLMtqNmhOpVKEw2HAbuWyVDabZWpqasXj/+M//iPpdJpPfOIT+P1+DMPgnHPO4dnPfjavf/3rNzWe2rvMuc+f/xG+edeGD9k3chp3zMj6uJ1nGAYej0c+6MeY044llUoxPT2NaZr9DkmIvnAmeKxEa82NN95IsVgkHA7j9/t54IEH2obCJBIJcrkcxx577IrHuvPOOznggAPw+Xxt27du3cp3vvMdHnroobZZxOvxl38Gtz8Oc4X27Vf8Dl5x9IYO2TeS7I0ZSfa6w7Ks5qkJ+aAfP0opotFoc8LG1NSUjN8UY8kZs7cWwWCQUqnEC17wAn72s58xOzvbHNd3ww03AHDWWWe17bNr1y723Xff5u2ZmRnuuOMOarVa24zhRx55pDmRaqOeeSD88lxoaPjgTfCfi7/aLX+y18pdrvI3qOQ07pgxDENO43aBZVkAUt0bY04LioWFBbLZbL/DEWLgOU2bL730UtLpNF/60pfI5XLcf//9XHTRRZx99tkcf/zxzcdfcskl7Lffflx++eXNbW9729vIZrNceOGFzULGDTfcwFVXXcUrX/lKYrHYpmI0DXs1jTO2tm///v2bOmzPSbI3ZqSy1x1utxvTNCXZG3M+n49AIEChUKBUKvU7HCHWZbk+e06F7V3vehfbtm3jmmuuAezTtdu2bevIxLStW7fyk5/8hB/84Ac8//nP58UvfjEnn3wyl156KbOzs+Tzeer1OtPT04RCobaZ7y9/+cu59tpr+cUvfsFRRx3FMcccw/nnn8/FF1/Ml770pU3H5jh2S/vtd1xnV/eGhRqXD/7jjjtO79ixo99h9F0ymWy+aERnpdNpyuUyW7ZskVN4Y0xrTSKRoFqtMj09vddmtEKIvdNaUy6XKRaLVCoVlFJ4vV78fn/zTEqvffyX8LFbdt8+5SD4ysu693xKqdu01sd14lhS2RszUtnrHsuyaDQa0mB5zDnjhJRSJJNJeb0JsQHOLN6pqSlmZmYIBAJUKhUSiQRzc3PNal8vnX1U++0/DNHy2JLsjRkZs9c9Mm5POEzTZHJyklqtRjqd7nc4Qgw1l8tFOBxmy5YtTE5OYhgG2WyWubk5kskkpVKpJ1+q9g/D5S3zRYbp/I2cXxgzUtnrntYWLMt1gxfjxbIswuEw2WwWwzCIRCL9DkmIoeZU+3w+H7VajWKxSKlUolwuYxhG8z6Px9O1GKYDu68/mIb8AgS793QdI5W9MWMYBlprSfi6xLIsFhYWpHoqALu1RDAYpFAoLNs0VgixMU61b2ZmhqmpqeaylfF4vKeneU/4wnBM1JBkb8zI+rjdJadyxVJO09hcLidLqgnRYUopLMticnKSLVu2MDEx0TzNOzs7SyKR6Ohp3ml/e3+9QhUO/zRcehMMcg1Fkr0xI+vjdpfb7cYwDEn2RJtIJNJcUq1YLPY7HCFGkmEY+P1+YrEYMzMzhEIharUaqVSKxx9/nFQqRblc3tTnX9iCfzt9z+2fuQ12PLaJ4LtMkr0x41T2JNnrDudbpiR7opVSiomJCSzLarboEUJ0j8vlIhQKsWXLFqampvD7/VQqFZLJJLOzs6TTaSqVyoY+C597KFx59p7b5wb4e5wke2PGqezJadzusSyLer0uLVhEG2dJNY/HQyqVki8EQvSIZVlEIpFm4uf1eimXyyQSCWZnZ8lkMutuDn3cfvDI+XBUywIdcUn2xKCQyl73ybg9sTdOwudyuUgmkx1ZfUAIsTbOmZeJiQm2bNlCNBptm9gxOztLNptd1+vyoInd1y/86eCO25Nkb8zIBI3uM00Tl8slyZ5YlmEYRKNRTNMkmUxKBViIPnBW5JicnGSfffZhcnISt9tNoVBoJn6ZTGbVU71HL1mMqjqgH62S7I0ZmaDRG16vl4WFBfk7i2WZpsnU1FRzlY1abQh6Nwgxopz+fdFotNm42e12UywWm6d6nbG2S9/T/+qp7ce6L9nDwNdBkr0xI5W93rAsC621VPfEXjkJn9a6uWa1EKK/nObM0WiUffbZp3mqt1wuk0wmm7N6S6USjUajrQ0LwL/f1p+4VyMraIwZpZSsotEDHo8HpRSVSgWv19vvcMSAcrlcRKNREokEyWSSqampZvVdCNFfzqler9eL1pqFhYXmih2lUqk5BvCIsMXvM14ayiQ9oBPt5V1lDMn6uN0nLVjEWnk8HqLRKLVajWQyKa9NIQbQ0skdsVgMv99PrVbjb47NMFmfJVKfx6zmBnLilVT2xpBU9nrDKf3XajVcLnmpib1zVgBIpVIkEonmBA4hxOBRSuHxeJpr8AZyNQpGGY8uUyrk2DWbw3IbeL1eLMvCsqy+V+ylsjeGpLLXG9KCRayH1+ttVvji8bhM2hBiSBimi7IRJGvG+GV6H152zSQ3/sn+sp9KpZrLthUKhb69riXZG0NS2esNl8uFaZqS7Ik1syyLWCyG1pp4PD6Qp4OEEO3C1u7rWhkkaz7+64HJ5uneQCBAvV4nk8kwNzfXNru3V4UXObc0hgzDkN5ePeL1eikWi2itmzOhhViJ2+0mFouRTCZJJBJMTk7KJB8hBthT9oFXHAXfvHv3tnKt/XRvOBymVqtRqVSoVCqUy2WKxSJKKdxud/N0r9vt7spnhVT2xpBU9nrHacEiFRqxHi6Xi6mpKVwuF6lUimJxgNdhEmLMKQUffj5871W7tzWW+Yh1uVwEAoFmP79YLEYwGERrTS6XazZz7sZrXip7Y0iSvd5pbcHijOETYi2cPnypVIp0Ok2j0SAYDPY7LCHEGvwuDn/IwBMiy9/fWvULhUI0Go1m1a9SqVAqlToaj1T2xpAzQUMSvu4zDAO32y3j9sSGOEur+Xw+stksmUym3yEJIfYisuT7/L/esvZ9nWbOTmuX6enp1XdaB0n2xpAzHkCSvd7wer1Uq1VZIUFsiFKKyclJAoEAhUKBVColr10hBtDSKt63fgfFDQ6Pd7vdmw+ohSR7Y0jWx+0tacEiOiESiRAOhymVStJ8WYgB9YUz2m8/MCBr5UqyN4ZkfdzecrvdGIYhyZ7YtGAwyMTEBAsLCyQSCakWCzFgnntIvyNYniR7Y0gqe70nS6eJTvH7/UxOTjabL8v/KyEGh6HgmJl+R7EnSfbGkFT2es/r9dJoNKS/oegIr9dLLBZDKUUymSSfz/c7JCHEAJNkbwxJZa/3nHF75XK5z5GIUeE0X/Z6vWSzWRnHJ4TYK0n2xpBU9npPWrCIbjAMg8nJSSKRCJVKhXg8LtVjIQbIPYl+R2CTZG8MSWWvPyzLolqtSpItOi4QCDA1NdVcU1dW3BCif1pXz3jn9XD3fP9icUiyN4aUUrKKRh94vV601lLdE13h8XiYnp7G4/GQTqdJp9PyGheiD07Yr/32Fb/rTxytJNkbU0opqTD1mLRgEd3mrLgRDAYpFovE43FqtVq/wxJirFx0SvvtygC8BCXZG1NS2es9Zy3ESqUif3vRNUopwuEw0WiUer1OPB6XiUFC9JBS8IFT+x1FO0n2xpSzPq7oLb/fT71elw9f0XVOexbTNEkmk2SzWfmSIcSYkmRvTEllrz+8Xi8ul4tCodDvUMQYcLlcxGIxAoEA+Xye+fl5FhYW+h2WEKLHOpLsKaWe1YnjiN6Ryl7/BAIBFhYW5ENX9IRSikgk0jZbV6p8QoyXdSd7SqmQUupApdQTnAvwoS7EJrpIKnv94/f7MQxDqnuipyzLYnp6Gr/fL1U+IcaMa60PVEqdBPwn8KSldwGSNQwZqez1j1IKv99PoVCgXq9jmma/QxJjwjAMJiYm8Pl8pNNp4vE4wWCQUCjUbLYuhBg966nsfRa4HjgTeC5wWstl51oOoJSaUUp9XSl17+LlCqXUAWsNQCn1Z0qpq5VSv1ZK3bN4jH9Zx+8gFkllr78CgQBaa6nuib5wqnwylk+I8bDmyh5Q0Fr/zXJ3KKUuXG1npZQHO1n8PXA0djXwP4EblFJP0VqvuJK3UuqZwBXA2Vrrmxe3vRV4F/B36/g9BPY3fK01Wmv5Rt8Hpmni9XopFotSVRF9YRgGkUgEr9crVT4hRtx6KnsPKKXcmzjO64FjgXdrrWta6zrwbuBQ4M0r7ajsd57/AD7mJHqLPrfavmJ5zpu5VPf6JxgM0mg0KJVK/Q5FjDGp8gkx+taT7P0Q+I5S6i+UUqcqpZ7tXID/u4b9zwL+oLV+0NmgtX4cuHvxvpWcDBwBfK91o9a6qrX+wTp+B7HIWR9Xxu31j8fjwe12k8+vWNQWouucKl/rjN10Ok29Xu93aEKIDljPadwvLv48fZn71lIeOhb7FO5SD2GPAVzJMxd/RpRSV2NPEqkD1wAXa62lNLJOUtkbDIFAgHQ6TaVSwbKsfocjxpxlWczMzJDL5SgUCpRKJUKhEIFAQE7tCrFBX7sT/vk5YPaxs/F6nvoW4JBlLocCv1rD/jEgt8z2LOBXSvlW2PfAxZ//hZ3cHQ28FvhL4Oq97aSUepNSaodSasf8/PwaQhwfzhu3VPb6y+fzSRsWMVCc5damp6exLItsNsv8/Lys+iLEOoQ87bd37OpPHI71JHsf1Fo/sszlYeA9XYrP4V38+R9a618BaK1/g93f7/lKqVOW20lr/Tmt9XFa6+Omp6e7HOJwcU7jSmWvv5RSBAIByuWyLFgvBorL5SIajTI1NQVAMpkkkUjI/1Mh1uB/PbH9drHanzgca072tNbfBVBK+ZRST1u8+Bbvu2ENh4gDoWW2h4HiKqdinYrgHUu237748/g1PL9oIZW9weH3+1FKSXVPDCRnAkckEqFarTI/P08mk5H3DiFW4HfDKQf1O4rd1nUGWSl1ETCPfdr2V8D84ra12AkcvMz2Q4A7V9n3nsWfS+Ot72W7WIVU9gaHaZr4fD6KxaJ8gIqB5FSgZ2Zmmg3B5+bmKBQK8h4ixBBYc5KklPpb4E3AvwLnLl4+CbxJKXXBGg5xFXCQUurglmNuAY4ErlzyXFuUUq2xfR87sTt2yTGPWfx561p/D2GTyt5gcZosF4vFfocixF45s3anp6dxu91kMhni8biM5xNiwK2nIvY64Ola6/dorb+8ePkH4OnYEyVW8yXsCt6HlFKuxWTuUuzZuJ9xHrS4LNtjwKedbVrrR7GTzLcqpQ5bfNz+2M2Ur1/jaWTRQiklq2gMELfbjcfjkUqJGAput5upqSkmJyfRWpNMJonH41QqlX6HJsRAuuin/X3+9SR7hcWkq83itlUHG2mtF4DnY1fo7gZ+hz1e77Qlq2fkgQywdO7Ku7Arid9XSt0D/A/wbeCl6/gdRAtZH3ewBAIB6vW6fGCKoeHz+ZienmZiYoJ6vU4ikZCkT4hF0ZYeIw9n+hcHrK/PXlAptZ/W+rHWjUqpA4HAWg6gtZ4FXr3KY34DRJfZXgc+uHgRHSCVvcHi9XoxTZNCoYDX6119ByEGgFIKv9/fHHeaz+dJJBJYlkUoFMLj8ax+ECFG0F8/Fb59z+7bD6fh4In+xLKeZO9rwK1Kqf8E7lvcdjj2KdyPdjgu0QNS2RssziD4bDZLtVrF7d7b6oRCDB7n/6/f76dYLJLL5YjH45L0ibF12FT77c//Gi4+rT+xrDnZ01r/i1JqAvhbdve9KwMf0Vr/v24EJ7pLKnuDx+/3N1cvmJjo01dAITahNekrFArk83ni8Ther5dQKCRfYsTYcBkw4YX04vyle+L9i2VdLUu01u8BprEnZTwdmNZaX9iNwET3SbI3eAzDwO/3UyqVpOoqhppSimAwyMzMDKFQiIWFBebn50kmkzKmT4yNS1oqeTt2wbuu708c6+5Pp7UuaK1vXbwUAJRS7+t8aKLb5DTuYHLasEiTZTEKDMMgFAq1JX2JRIL5+XlKpZJ84RQjLbpkIdhv3Q2Fhd7HseJp3MXJF1mtdUYp9boVHvpq4AMdjUx0nVT2BpPL5cKyLIrFIsFgUBagFyPBSfqCwSDFYpFCoUAqlcI0TYLBYHOdaCFGyQn7w2uPha/utG9roNaHGstqY/Z2APcCz8buk7c3kjEMIansDa5gMEgikaBUKuH3+/sdjhAd44zpc9aEzufzZDIZcrkcfr+fQCCAaZr9DlOIjjAUfOBUuPpeyPZx9MJqyd7p7F6X9mda61OXe5BSSpoaD6HWVTTkG/VgsSwLl8tFoVCQZE+MLK/Xi9frZWFhoTmZo1Ao4PP5CAQCMplDiA5ZMdnTWv+65ebfrfDQle4TA0rWxx1sgUCATCZDuVyWvntipHk8HjweD6FQiEKhQLFYpFgsYlkWfr8fr9crwxmE2IT1tF5pW39WKRXAXhHjnqX3ieHQWtmT0yaDx+lXlslk8Hg8Un0VI8/lchGJRAiFQnuM63OqffJeJcT6rfnTQyn1NqXUfUqpE5RSLuBm4CrgTqXUn3ctQtE1UtkbbEopIpEI9XqdXC63+g5CjAjDMJptW6LRKG63m3w+z+zsLMlkknK5LO9bQqzDelbQeCVwptb6TqXUmcBWYBvgBj6FvU6tGCKtlT0xmDweD4FAoDl2T8YwiXGilGqO66vX683Tu8lkEtM08fv9+P1+qfYJsYr1JHsVrfWdi9fPAr6ltd4JoJQqdzwy0XVS2RsOoVCIcrlMOp0mFovJ2CUxlkzTbLZuqVQqFAoFcrkc+Xy+ObbPsix5fQixjPUkexY0x+q9GHhVy32SLQwh501Rkr3BZhgG4XCYVCpFsVgkEAj0OyQh+qa12ler1ZrVvnK53Bzb5/P5pAouRIv1JHv3K6W+CuwDpIHrlFIGsJ0NrMQh+s+p7Mlp3MHn8/kolUpks1m8Xq+cthICe0JHOBwmFApRqVSakzry+Txut7uZ+MnrRYy79SRpbwOSQB44S2vdwE70zgc+04XYRJdJZW+4hMNhADKZTJ8jEWKwONW+aDTKli1biEQiKKXIZrPMzs6SSCQoFovyxVYMhEK198+5ntYrOezErnXbd4DvLM7OFUNIVtEYHi6Xi1AoRDabld57QuyFYRjNFTpqtRqlUolSqUQ6nW4mhT6fT8b3ib458T/hzCPgYy/o3XN26vTrDzt0HNFjsj7ucHFWFchkMpKkC7EK5wvSzMwMsVgMv99PpVIhmUwyOztLOp2WNi6iJ6wlIwmuugf+lO3d869YkVNKnQ/8SWt9hVLqJys8dFtnwxK9IpW94eL03ovH4+Tz+eapXSHEypxVOsLhMJVKhXK5TLlcplgsYhgGlmVJxU90zb++EF51Vfu2Uq13z7/a6dfXAPcCVwBHAtfu5XFHdDIo0TtS2Rs+Tu+9fD4vsw6FWKfW2bxa67bEr1Qqtd0vy7SJTnnmgfDI+XDaV+CBVO+ff7W1cU9ouflzrfUblnucUurKjkYlesYwDOr1er/DEOskvfeE2Lylid/CwgKlUmnZxM+yLFmyUAyt9UzQOGsj94nBJpW94SS994ToLKUUlmVhWVYz8Wut+IFdVXeSP5dL5iWK4bHm/61KqeOxGyn/Smt92eK21wKG1vrLXYpPdJmM2RtePp+PYrFILpeT3ntCdFBr4heJRFhYWGie7s1ms2SzWVwuF5Zl4fV68Xg8Ul0XA209X03+bvHxrYndHcBHlFKTWuuPdzQy0RNS2RtukUiE+fl5MpkM0Wi03+EIMZKcyR2hUIh6vU65XG5r4uwkh87pXvniJQbNepK9JwAnLjZTBkBrfadS6gzgBkCSvSFkGAZaa7TW8s10CEnvPSF6yzTNZh8/Z4JH6yQPALfb3awMStVPDIL1JHu11kTPobVeUEpJaWhIta6iIW9IwykQCFAqlchkMtI2Qogeap3AEYlEqFarzeTPWbZNKYXH42kmfzJ7XvTDeqYW+ZRSRy3dqJQ6BvB3LiTRS05iIOP2hpfTe69er5NOp/sdjhBjy+12EwwGmZqaYp999mFqaopAIECj0SCbzTI/P8/jjz/enFg1qJ0QHn30UV7ykpdw5JFHsnXrVs4880wAdu3axQtf+EL5QjmE1lPZ+zhwi1Lq28D9i9u2Ai8D3tzpwERvOK0EZNzecHOaxToDx0OhUL9DEmKstU7yAKjX682qn9PiBeyhGE7lz+PxDMR4v7e85S3kcjnuuusuSqUS27Zt46qrruKCCy7oWmXytttu44ILLiCRSFCtVjnjjDP4wAc+sOahKVdeeSWf+MQnKBQKpFIpotEo559/Pq997Wu7Eu+wWXNlT2v9FeAdwHOAixYvzwLerrX+ahdiEz0glb3REQwG8fv95HK55geJEGIwmKaJ3+9ncnKSLVu2MD09TTgcxuVyUS6XSaVSzM7OMjc3Rzqd7mvl78Ybb+TEE09srjO8c+dOPvShD3H99ddz0kkndfz57rvvPk499VTOPPNMfvvb33LLLbdw3XXX8YY3LNvadw8f+9jHuPjii/nGN77Bbbfdxr333svWrVv58Y9/3PFYh9W6GgVprb8AfEEpNb14e74rUYmekcreaGk9nWuaJh6Pp98hCSGW4Xa7m1UyrTW1Wq1Z9XOWcYPdlT/n0ov+fplMpq2i5vP5uPnmm7v23BdddBHRaJS3v/3tAExMTHDhhRfyile8ggsuuIDjjz9+r/s+/PDD/P3f/z033XQTBxxwAGD/bT/ykY/w2GOPdSXeYbTuduBKqSOAY7XW80qpWBdiEj0klb3RopRicnIS0zRJJpPUaj1cfFEIsSFKqeZ4v2g02qz8RSKRZuUvnU4zNzfH7OwsyWSSfD7PwsJCR7+oX3PNNWzbZi91/9nPfpZt27bxute9DqBriV6tVuPqq6/mlFNOaRsLeNpppwH26dmVfPWrX2ViYmKPhHC//fbjuOOO63zAQ2o9TZVDwOXAC4GHgCcC/6KUOgA4W2ud6U6Iopuksjd6DMMgGo0Sj8dJJpPEYjFZ5kmIIeIkf263u7k6TrVaZWFhgYWFBarVarPNi/PY1urfRl/v27dvZ/v27SilOO+887jooovWfQwnWVzNHXfcAcCDDz5IoVDgkEMOabt/amqKUCjEzp07VzzOz3/+cw4++GCuvPJKPv7xjzM/P080GuWNb3wj55577rrjH1XrSdUvAcrAi4B/BNBan6uU+hvgw8CbOh+e6Dap7I0ml8vF5OQkyWSyOVhZZtAJMbyWJn/1er0tAXRavcDuU7/OPm63u2evfyeJW6t4PA6w7KSycDhMIpFYcf9HH32Uhx9+mI985CN8+9vfZmZmhiuvvJJXvepV7Nq1i/e+973rimdUrSf9/zPsCt61QMXZqLX+FHBYpwMTvaGUklU0RpSz1FOlUiGbzfY7HCFEB5mmidfrJRwOE4vF2GeffYjFYs1JH5VKhUwmQzwe5/HHHycej5PJZCgWiyM1vKNcLlMoFPjwhz/MPvvsg2EYnH322bz0pS/lgx/8YHPs47hbT2VPaa33NjUo3IlgRH/I+rijy+/3U6vVyOfzuFyuZlVACDFanObNrZOy6vV687RvtVqlWCw2v9gbhtGs+jlVwH60fYnF7KH/uVxuj/uy2SzHHnvsivs7FcGlp4+f8pSncNVVV3H33XfL2D3Wl+yZSqnjtda3tm5USp0JVDsbluglqeyNtnA4TK1WI5PJNKsBQojRZ5omPp8Pn88H7J716yR/S0//Ogkg2GMEa7XauidmrHfM3qGHHkogEODhhx9uuz+RSJDL5VZN9o444gjuuOOOPQoWTuI6yIWMr+yEf35Ob55rPf+KFwM3KqV+BBymlPoScBTwFOAlXYhN9Igke6NvcnKSeDxOKpUiFovJkk1CjKHWiR8OrXUz+XMuYJ8enZuba9vH7Xavmjytd8yey+Vi+/bt/OxnP2tbtvOGG24A4Kyzzmp7/K5du9h3332bt8844wwuu+wydu7cycknn9zc/tvf/hafz8fRRx+9rni6rfWj9scP9S7ZW09T5e8BLwUiQBQ4G3vCxula6+u6E57oBTmNO/qUUkSjUQzDIJlMDuwyTUKI3nJO/wYCASYmJpiengbsISATExP4/fZqqMVikXQ63ZwFPD8/TyqVIp/PU6lUNvWe8k//9E8kEgk+9alPAXafv/e///2cc845bS1VLrnkEvbbbz8uv/zy5rZXvvKVHH/88bzvfe9rVihvvPFGrrjiCt773vcO3NCVv3rq7ut/zMJsvjfPu+ZkTyl1LPBHrfWztNaBxcuztdY/6mJ8ogeksjceTNMkGo3SaDRIJpPyby6EaNPaZ+/zn/88z3zmM/n1r39NLBbjox/9KKeffjo/+pH9kX/aaafxrGc9i3g8TiKRYHZ2lscff5xEIkE2m6VYLFKtVtf0PnPYYYfxk5/8fi7xTgAAIABJREFUhCuuuIKjjz6a448/nuc973l88YtfbHvc9PQ0oVCIaDTa3GaaJtdeey2HHXYYRx99NIcffjhvfetb+dSnPjWQM3FPOaj99rfu7s3zqrW+4SulGsBlWutXdzek7jjuuOP0jh07+h3GQEqn01QqFbZs2dLvUEQPlMtlkskkXq+XyclJackihNiwRqPRHN/nnAau1WptSZ7L5cLlcuF2u5vXXS7X2L73HPSJ3df/6qnwvmct/zil1G1a647MLlnPmL1bhjXREyuTyt548Xq9TExMkE6nSSQSzdO7QgixXoZhYFkWlmU1t2mtm30Aa7VaMxGsVCrNzxqlFKZpjmUS+J6T4YM39fY515PsPayU8mqty0vvUEp9Rmv95g7GJXpIxuyNH7/fj1KKdDpNPB5namqqL20XhBCjRynVTNxaObOBWxPAWq3WHAfoaE38Wi+j+KW0V3WW9SR71wLXKKW+BjwKtI7GfHpHoxI91bqKxii+mMTyfD4fhmGQSqWIx+NEo1GZpSuE6JrlZgPDnkmgc2mtBIJdmFguCTRNc2irgV+43a70mV3+6F1PsueMlHzeMvfJOcAhJuvjji/LspiamiKZTDZP6bY2ZRVCiG7bWxIILJsElsvlPc5GOUnf0iRwEBPBqK/99s45eMo+3X3OdY3ZA85ZZrsC/qsz4Yh+aK3syam88eN2u9sSvsnJSWm8LIQYCMudDgb786pWq1Gv19sSwVKp1JYIto4NXO5nPxLBFz0J3nn97tsLPeiEtWqyp5Q6EHgR8A0gr7XeY1VipdR7uhCb6BGp7AmXy0UsFiORSJBKpYhEIs3+WkIIMWgMw9jrWYh6vb5HEliv1/c4LQysmAh2a1hTwAMn7Ae/eqwrh1/WismeUuppwI+wGykDvFsp9Uyt9R9aH6e1vqFL8YkecL7ZSLI33gzDIBaLkUwmSafTNBoNgsFgv8MSQoh1cU7fLpcMOhXB1qpgvV5f9tSwYRjNY7Umgc71QTs9vJLVKnsXA98CPg1YwHsXL3/d5bhEDznfXmRGrnBW2kin02SzWer1OpFIZPUdhRBiCDgVwb0lgq1VQef63qqCzoSR1oSw9TJIEx5XS/b2xV4OTQMopV4DXL/yLmLYSGVPtFJKMTk5iWma5PN5Go0GExMTQ/UtVggh1sswDAzD2GtXgtZxgq0JYbVapVwuL5sM7i0RVNoEbUCP3ldXS/bSuiV6rXVeKVVZ+iCl1Eu11ld3PDrRE60TNIRwhMNhDMMgm802Ez6ZwCOEGFcrjRME2iqBSy8LCwttn7FWGabq0FAm+bRJyr9nQthJqyV7y80RWS4jOB+QZG9IyQQNsTfBYBDTNEmn08zPzzMxMSEzdYUQYhmrJWmtp4mr7gZFo46p6yi19+pgp6yW7D1FKfWTJdu2LbetgzGJPpBVNMTe+Hw+3G43qVSKZDJJIBAgHA7LaV0hhFiH1tPEdReUFof0BSIwM2Nfb00IO2ktffaWvqP/ZpltYsjJ+rhiJU5rllwuRz6fZ2FhgYmJCVlxQwghOmi1cYMbtVqyd4fW+tTVDqKUktYrQ84wjI5/kxCjRSlFOBzGsqzmmrrhcJhAINDv0IQQQqxgtXnBr1njcdb6uI5QSn1AKaWVUn/Zy+cdZZZlsbCwINU9sSrLspiensayLDKZDMlkUoYACCHEAFsx2dNar6m/81ofp5SaUUp9XSl17+LlCqXUAWvZt+UYBwAXrGcfsTqPx4PWmoWFhX6HIoaAYRhEo1EikQiVSoX5+XkqlT0m6gshhFjFl+7o/nP0rOOfUsqD3aPPAxwNHAUUgBuUUutp0/9BYOkEEbFJlmWhlJIPbLEugUCAWCyGYRgkEgkymYxUh4UQYhWelkm7378fsl3+6O1le+fXA8cC79Za17TWdeDdwKHAm9dygMXl204CPtm1KMeUUgq32y3Jnlg3t9tNLBYjEAhQKBSIx+NSIRZCiBX85ZIeJl/6TXefr5fJ3lnAH7TWDzobtNaPA3cv3rcWH8Verk0yki6wLItqtSrjr8S6KaWIRCJEo1EajQbxeJx0Oi2TfoQQYhnPP7T99kOp7j5fL5O9Y4GHltn+EPDk1XZWSr0M8AGXdzgusciyLACp7okN83q9TE9PEwwGKZVKzM3Nkc/n5dSuEEIs8cH/396dh0dZnosf/z6zZt+TCSqbyqIIJwW0pVA3LLVqqWIVrBu2itpWa1GqPz0ii0ut9kjVWuqGVkUUpYAWKyBolXMOiKeIuNJWBRXIAmTPZJb798cszmQmyYRMMlnuz3W912Sed5n7zZuZufNs76nd91rdmewVAbVxymuADGNMems7GmPswN3A9dKBbw1jzCxjzFZjzNaKiooOB9zf2O12LBaLNsGpTrFYLOTk5IRH7NbU1FBRUUFTU1OqQ1NKqR4jLZGZjpOkO5O9zrgaeF9E3urITiLysIiMF5HxxcXFXRRa32GMweFwaM2eSgqbzUZBQQGFhYUA7N+/n6qqKrxeb4ojU0qp/qU7k71KIDtOeQ7QICKN8XYyxuQB/4/AYA7VxZxOJ16vV7+QVdKE5uXLzc3F4/FQUVFBdXW19g1VSqmg9fE6uSVRdyZ724EhccqHAu+1sd+3AC+w3BizzRizDXg0uG5BsGxuUiPtx0L99rQpVyWTMYbMzExKSkpIT0+nvr6e8vJyGhoatD+fUqrfq3HDzv1dd/zuTPZWAIONMUNCBcYYF3AM8GLkhsYYlzHGAiAifxORgSJSFlqAy4Obzg2WLeiWM+gHbDYbVqtVm3JVl7BYLOTl5VFcXIzNZuPgwYNUVFRo0qeU6neOPyz6+SdVXfda3ZnsPUGgBu9uY4wtmMz9hsBo3D+GNjLGTAS+Av7QjbGpCE6nU5M91aVCc/Pl5+djjOHgwYOUl5dTX1+vSZ9Sql8YlAvDC7vntbot2RORZuC7gI/A3HofEuivd6qI1EVsWgdUA3taHiN4u7V4zbjjuzT4fsbpdOL3+/F4PKkORfVx6enpFBcXU1BQgNVqpbq6WqdrUUr1G0flf/1zZUPXvU43DvwFEdkH/Lidbd4FClpZVw6UxVunksfhcACB+fbsdnuKo1H9QVpaGmlpabjdburq6qipqaGuro7MzEwyMzOxWHrLxAFKKXVo5r4Ol4wBY5J/bP0EVTGsVqveOk2lhNPppLCwkKKiIhwOB7W1tZSXl1NTU6Ojd5VSfc6oFrPCNXfRTYc02VNxORwOmpubtSlNpYTD4aCgoCA8MXNdXR379u2jurpauxcopfqMK8ZGP9+0u2teR5M9FZfT6UREdAoWlVJ2u538/PzwlC0NDQ1UVFRQWVmpI3iVUr1ey7toLN3RNa+jyZ6Ky+FwYIzRplzVI9hsNvLy8nC5XOTk5OD3+zl48CD79u2jpqZGJwFXSvVa3z/665/dXfRRpsmeistisWC327VmT/UoFouFrKwsSkpKKCwsxOl0hidorqqqorGxUWv7lFK9ygXHdf1rdOtoXNW7OJ1Oamtr8fv9OhpS9ThOpxOn04nP56OxsZH6+noOHDiAxWIhIyODjIwMbDb9iFNKKf0kVK0KJXtut5v09PRUh6NUXFarlaysLLKysnC73dTX11NfX09dXR0Oh4P09HTS0tKwWq2pDlUppVJCkz3VKrvdjjGG5uZmTfZUr9Cytq+xsZHq6mqqq6txOp3hxE9rqpVS/Ykme6pVxhi9dZrqlSJr+7xebzjxO3jwYPjvOjSRsyZ+Sqm+TpM91San00lTUxM+n0+bwVSvZLPZyM7OJjs7G4/HE078mpqaMMaEk760tDRMV0xdr5RSKabJnmqT0+kEArdOy8jISHE0SnWO3W7HbreTk5NDc3NzOOlrbGzEGIPD4QgnfvrPjVKqr9BkT7XJZrNhtVo12VN9jsPhwOFwhBM/t9tNU1NTuI+f3W4PN/eG+q8qpVRvpMmealeoKVepvijUh8/pdJKTk4PX66WpqSk8sreurg6LxUJaWlp4O+3np5TqTTTZU+1yOBw0NDTg8Xiw2+2pDkepLmWz2cKDO/x+P263O1zr19DQgDEmXOvndDq11k8p1eNpsqfaFdlvT5M91Z9YLBbS09NJT09HRPB4POFav9raWmpra8N9/ULJn81m0+RPKdWjaLKn2mW1WrHZbLjdbrKyslIdjlIpEUrqHA4HAH6/P9zXz+12U1NTAwQSxJbJn1JKpZJ+CqmEOJ1OGhoaEBGttVAKwv340tLSAPD5fFHJX6ifq9VqxW63hxNFbfZVSnU3TfZUQkI3nG9ubg436yqlvma1WsNNvgBer5fm5ubwEkr+Qn3+Qsmfw+HQAR9KKQD+vgt8frAm+SNBkz2VEIfDgTEGt9utyZ5SCbDZbNhstvCURT6fD4/HE07+QiN9Q9tG1vxpvz+l+o+Wed3rn8Pkocl9DU32VEIsFgt2u53m5uZUh6JUr2S1WrFareFm39CAj8iav4aGBuDr2r/IRRNApfqmcYdFP/+iJvmvocmeSpjT6aS2tha/36/NTkp1UssBHxBo+vV4POGlsbGR+vr68PaaACrV92TY4UfHwAsfdt1raLKnEhZK9pqbm8O1E0qp5Ak1/Yb6/UF0Atjc3BweKAWBBNBms0Ulf3a7Xf8ZU6qXSe/iWc30E0ElLDSK0O12pzoUpfqNUPKXk5NDUVERpaWllJSUkJeXR2ZmJhaLBbfbTXV1NVVVVezdu5d9+/ZRVVVFTU0NjY2NeDyecIKoVHt2797NWWedxTHHHMPw4cOZNm0aAHv27OH000/v0bXJ3/nOdzDG8Nlnn6U6lB5Fa/ZUwkK3ldJkT6nUCdXmtZy/z+fzhWsBQ4/19fVRtYChOTNDtYChpSd/eavu97Of/Yza2lref/99GhsbKSsrY8WKFcyePbvLJtZ/5513mD17NlVVVXg8Hn7wgx9w++23d6gV6cUXX+Stt97qkvh6O63ZUx3idDrxer34fL5Uh6KUimC1WnE6nWRlZZGXl0dxcTGlpaUUFxeTn59PZmYmNpsNr9dLXV0dBw4coKKigj179lBeXs7+/fupqamhoaGB5uZm/H5/qk9Jpcibb77JhAkTsFgsZGZmsn37du6++27WrVvHxIkTk/56O3fu5JRTTmHatGns2LGDzZs38+qrr3LZZZclfIzm5mZuuukmzjjjjKTH192e2p78Y2rNnuqQUGdyt9sdnlJCKdUzRQ7qiOwHKCJ4vd6Yxe12RzX3WiwWbDZbuEYwtFitVu0X2IdVV1dH1ailp6ezadOmLrsbzLx58ygoKODaa68FIC8vj7lz53L++ecze/Zsjj/++HaP8Yc//IHjjz+e4cOHs2bNmi6Jsys5It5OO/dDTZIb0PTdqjrEbrdjtVq1KVepXiyUBKanp5OdnU1+fn64JrCkpISCggJycnJIT0/HGENzczO1tbXh2sBQv8DKykoOHjxIXV0djY2NWiPYy61evZqysjIAFi9eTFlZGZdccglAlyV6Xq+XVatWcdJJJ0V1Jzj11FOBQNNse/bv388999zDXXfd1SUxdoepI6Kf1yV5ljOt2VMd5nA4cLvdOgWLUn1Ma/0B4evawFDfwMjawND8gCEWiyU8r2BkzWCoTPsI9kxTp05l6tSpGGO46qqrmDdvXoePEUoW27Nt2zYA/v3vf1NfX8/QodGzCBcWFpKdnc327e23aS5YsICLLrqIwYMHdzjenqKsFPLS4GDgRjtJn2tPkz3VYZmZmTQ2NlJXV0dOTk6qw1FKdYPIJuGWIhPBUDIYemzZNAxfTzDd2qL/RPZeoSQuUZWVlQBkZ2fHrMvJyaGqqqrN/Xfu3Mnzzz/Phx924SR13cRh/frne/8nucfWZE91mMPhIDMzk/r6etLT07tsdJZSqndoKxEEYpLA0OLxeGhqaopJBiNrBlsmgVo7qCLdeOON3HTTTeTm5qY6lE4rzYTywBzq1GozruoJsrOzaWxspLq6mqKiolSHo5TqwUIJWuTdQiJFJoAtl9b6AcZLCENlkY+qZwt9f9TW1sasq6mpYcyYMa3u++abb7Jjxw6ee+65LouvO915Kpy1rGuOrcmeOiQWi4WcnBwOHjxIQ0ODjsxVSh2yULLWGhGJSQL9fn+7CWFobsGWSWDLMk0Kk6ejffaOPPJIMjMzYyZBrqqqora2ts1kb926dfh8vqjRunv37gXgjDPOwOFwcOedd/aa6Vi6srJakz11yDIyMmhoaKCmpoa0tDT9wFRKdYm2Bo6EhBLCyCQw9LPf78fj8eDz+eLeScQYE5P8xUsINTFsX0f77NlsNqZOncobb7yBiISb5zdu3AjAueeeG7X9nj17GDBgABAYmLFgwYKo9fPmzWP+/PmsWbOGIUOGHOJZ9D2a7KlOyc3NpbKykpqaGvLy8lIdjlKqnwolhO3x+/1RSWDkz5F9CVubOD6UGLaWCLYs076F7Zs/fz7jxo3jwQcf5JprrqG6upqFCxcyY8aMqFq7u+66i5tvvplly5Yxffr0FEbc+2iypzrFbreTmZlJXV0dGRkZrfbJUUqpniCUhLWXGIpIOBlsmRxGlnk8Hvx+f6v3Ho5MDttKCHtCgrh69Wrmzp0LBObZW7lyJffddx+nnHIKc+bMYd26dezatQv4url2y5Ytnf7cHzZsGBs2bOD6669n8eLF4dul3XHHHVHbFRcXk52dTUFBQcwx1qxZw8033xzTjNvRmsa+yvSXm2OPHz9etm7dmuow+iQRoby8HIvFQlFRkf4nq5Tqd1omgq0toaSxre/eeAlgyyVegqifvb3bjnI489nAz8cWw98uNO+IyPhkHFtr9lSnGWPIzc1l//791NfXk5WVleqQlFKqW3W0P1/LmsN4S2ibUO1he3cnaS0JTORnTRT7Nk32VFKkpaWRlpZGbW0t6enpbY6sU0qp/i40Urgjn5UiEpMktvXc4/GEn7fXihdK/uIlhIk+qp5Lkz2VNDk5OVRUVFBTU0N+fn6qw1FKqT4lVAN3KIlVZNIXLzFsuT40cjlUlkiXr5YJYLyksL0yrWXsGprsqaSx2WxkZWVRW1tLRkYGTqcz1SEppZTi65rEQ9UyGUz0sWXSmGisbSWDLRPDls81eYylyZ5KqqysrPCdNYqLi/VNppRSfUBnk0X4uhm6ZY1hIj+HksaOJo6h2NtLBhNd13K73kKTPZVUocEaVVVV1NXVxb25tVJKqf4nspYtGf26WyaE8Z63VRavvKMzlCSSFCaaRHo9BqsYBAOS3IoSTfZU0jmdTtLT06mrqyM9PT2hiU6VUkqpjkhGbWNL8ZK/1hLG9pZ4iWhbag9CXnAu70x3Uk9Lkz3VNXJycnC73VRXV1NYWJjqcJRSSql2dXUfv7aSw30+odYiGAS3LblzIGuyp7qE1WolOzub6upqGhsbSU9PT3VISimlVEq1lUzandAc7Aa4/UByX7f39C5UvU5GRgZ2u52ampp2q6+VUkqp/szehRmZJnuqy4QGa/j9fiorK1u9sbhSSinV3w3vwh5PmuypLuVwOCgoKAgnfF6vN9UhKaWUUj2OMXDdN7vm2JrsqS7ndDopLCxERKisrMTj8aQ6JKWUUqrH6aqhIZrsqW5ht9spKirCYrFQWVmJ253kceVKKaWUikuTPdVtbDYbhYWF2Gw29u/fT2NjY6pDUkoppfo8TfZUt7JarRQWFmK32zlw4AD19fWpDkkppZTq0zTZU93OYrFQWFhIWloa1dXV1NbWpjokpZRSqs/SZE+lhDGG/Px8MjIyqK2tpbq6OtUhKaWUUin17YFdc1y9g4ZKGWMMeXl5WCwW6urq8Pv95OXldemtapRSSqme6oTDYe2F8O4+mJ7E43ZrzZ4xpsQY84wx5uPg8oIx5ogE9htgjJlvjNlujNlhjPnIGLPCGDO6O+JWXSsnJ4ecnBwaGxvZv38/Ism9J6BSSinVW4wogvNHJfeY3ZbsGWMcwDrAAYwCjgXqgY3GmKx2dr8NuAA4U0SOA8oAH7BZE76+ISsri7y8PNxuN+Xl5TpSVymllEqS7qzZuxQYA9woIl4R8QE3AkcCVyew/29FZDeAiDQBNwHpwKwuild1s4yMjPBcfAcOHKCqqkonYFZKKaU6qTuTvXOBXSLy71CBiOwFPgiua8svgMdblH0VfMxPWoQq5RwOB0VFReTm5uLxeKioqKC6uhq/35/q0JRSSqleqTuTvTHAp3HKPwXabIoN1gS2/LYfHnx8vbX9jDGzjDFbjTFbKyoqOhKrSiFjDJmZmZSUlJCZmUl9fT3l5eXU19drfz6llFKqg7oz2SsC4k2oVgNkGGPSO3i8WcD7wFOtbSAiD4vIeBEZX1xc3MHDq1SzWCzk5uZSXFyM3W6nurqayspKmpubUx2aUkop1Wv0ynn2jDGTCYxKPl9E9CarfZzdbqewsJD8/Hz8fj+VlZUcOHAAn8+X6tCUUkqpHq8759mrBLLjlOcADSKS0PBLY8x/AE8CU0XkgyTGp3q49PR00tLSqKuro66ujqamJrKyssjIyMBqtaY6PKWUUqpH6s5kbzswMk75UOC9RA5gjBkDrARmiMh/JzE21UsYY8jOziY9PZ3a2trwkpaWRmZmJk6nM9UhKqWUUj1KdzbjrgAGG2OGhAqMMS7gGODFyA2NMS5jjKVF2RhgFXCxiLwVLBtgjPlTF8eteiCbzUZ+fj4lJSVkZWXR3NxMVVUV5eXl4btxKKWUUqp7k70nCNTg3W2MsQWTud8QGI37x9BGxpiJBKZV+UNE2WjgNeBVYIgx5iJjzEUE+u2N6LYzUD2OzWYjJycHl8tFfn4+VquVmpoa9u3bx4EDB3C7tUunUkqp/q3bmnFFpNkY813gPgJz6wmwAzhVROoiNq0DqoE9EWXzCYzmvTK4RHqjy4JWvYYxhvT0dNLT0/F6vdTX19PY2EhjYyM2m43MzEzS09OxWHrlmCSllFLqkJn+Mm/Z+PHjZevWrakOQ3UjEaGxsZGGhgaam5sxxuBwOEhLSyMtLU0HdSillOqxjDHviMj4ZByrOwdoKNWtjDFkZGSQkZGBx+OhsbGRpqYmqqurqa6uxm6343Q6SUtLw263Y4xJdchKKaVU0mmyp/oFu92O3W4nJycHr9dLU1MTbreb+vp66urqsFgspKWl4XQ6cTqd2tyrlFKqz9BvNNXv2Gw2srKyKCwsDA/scDqdNDU1ceDAAfbt20dVVRW1tbW43W69RZtSql/ZvXs3Z511FscccwzDhw9n2rRpAOzZs4fTTz9dW0F6Ia3ZU/2axWIJD+wQETweT7jWr7Y2cHc/Yww2mw2HwxFetL+fUqqv+tnPfkZtbS3vv/8+jY2NlJWVsWLFCmbPno3dbu+S13znnXeYPXs2VVVVeDwefvCDH3D77beTlpbW5n7btm3jD3/4A2+99RY2mw2fz8dpp53Grbfeit4m9Wtas6dUUGgAR05ODsXFxZSWllJQUEBWVhYWi4WGhoZwzV9oapf6+no8Ho/W/iml+ow333yTCRMmYLFYyMzMZPv27dx9992sW7eOiRMnJv31du7cySmnnMK0adPYsWMHmzdv5tVXX+Wyyy5rd98ZM2awf/9+tm7dynvvvce6detYu3YtEydOpLExoRtz9Qua7CnVilA/vuzsbAoLCyktLaW4uJjc3FwcDgfNzc1UV1dTUVHB3r17qaio4ODBg9TV1eF2u3ViZ6VUr1RdXR1Vo5aens6mTZsYNmxYl7zevHnzKCgo4NprrwUgLy+PuXPnsmzZMt5+++1297/77rvJzMwE4PDDD2fOnDns3LmTNWvWdEm8vZEme0olyBiD3W4nMzOT/Px8XC5XuM9fZmYmFosFt9tNTU0NVVVV7N27N9z/r6amhsbGRq0FVEr1WKtXr6asrAyAxYsXU1ZWxiWXXAIE+jp3Ba/Xy6pVqzjppJOi+gKeeuqpALz44out7QrA9u3bOfroo6PKDjvsMAAOHDiQ5Gh7L+2zp1QnWK3WcJ+/EL/fj8fjwePx4PV68Xg81NfXh5M8YwxWqxWbzRZ+DC0Wi0U7PyulUmLq1KlMnToVYwxXXXUV8+bN6/AxQslie7Zt2wbAv//9b+rr6xk6dGjU+sLCQrKzs9m+fXubx3E4HDFln3zyCcYYTjzxxASj7vs02VMqySwWS3gKlxARwev1xiwtR/tGJoKhxWq1hhdNBJVSPVkoiUtUZWUlANnZ2THrcnJyqKqq6tDxfD4fjz32GD/96U8ZPnx4h/btyzTZU6obhJqA441k8/l8eL3e8GNriSAQlfjFW3R+QKVUf7Zw4ULsdjuLFi1KdSg9iiZ7SqVYKFFrSUTw+/3hRDByCU0R0zIZDNUMhhK/lj+HHrWGUCnVExQVFQGEp7qKVFNTw5gxYxI+1pIlS3j++ed5/fXXwwM2VIAme0r1UJGJW2taJoE+nw+/399mQgiBpubIBLC156FFKaUS0dE+e0ceeSSZmZl89tlnUetDE9snmuw99dRT/O53v2PDhg2UlJR0KOb+QJM9pXqx9pJBIJz8tfbo8Xjw+/2tThVjjImbALZcWm6nlOp/Otpnz2azMXXqVN544w1EJNzqsHHjRgDOPffcqO337NnDgAEDosqefvpp7r77btavX09paSkAL7/8Ml999RWzZs061FPpUzTZU6qPSzT5CjUbh5ZQQthyaS85bPm6kUlgez9HPiql+of58+czbtw4HnzwQa655hqqq6tZuHAhM2bM4Pjjjw9vd9ddd3HzzTezbNkypk+fDsAzzzzDFVdcwcKFC1m/fn142zfffDMmKezPNNlTSgGJNRtHEpGYBDG0RJaHfg7NMRgqa09rSWCiP2viqFTHrF69mrlz5wKBefZWrlzJfffdxymnnMKcOXNYt24du3btAr5urt2yZUvc6U86YtiwYWzYsIHrr7+exYsXh2+Xdscdd0RtV1xcTHZ2NgUFBeGya665hqa076YAAAAgAElEQVSmJubMmRNz3Ntuu61TcfUlpr9M8Dp+/HjZunVrqsNQSkHcZDDyMV5ZvG0SFS8RjJcUHuqilFLJZox5R0TGJ+NYWrPXgtvtZv/+/dTW1uLz+VIdjlKqDa0lfO0lgt35T25HksFEt9UEs39xOBwUFRWRm5ub6lBUL6XJXgS3282uXbvIz89nyJAh2O12/VBVqp9pmQi297yt8kNNRjuirc+o9j6/OrNvMvZX7RMRGhsb+eKLL3A6nVH3rFUqUZrsRdi/fz/5+fnheX+UUv1PyySlK5OWRBPJttYdyj4h7Q2ySZbOJo6Hsl1XbdvdSawxhoyMDIqKiqioqGDgwIHd+vqqb9BkL0JtbS1DhgxJdRhKqX6iOxPLtnQmYUykljJZ20D3JaiJOpRrdij7pKenU15eTk1NTdzjdPTnZGzX2W1V99FkL4LP54t7OyullOrLenNzbLL7Z3b19oe6j9Vqxev1Ul9f36nj9AQdSRiT8byr9umusmTQZK+Fnv7BppRS6mud6ZfY21it1rhzx0UmfR39uTvWdWbbQ3ker6xljXAi+xzqNj2RJntKKaVUL9ZW06nqXl0xoCsZNNlTSimllEqCntpPUW9gqVpVXl6Ow+Fg0KBBPa5TdG9x5plnUlpaGr47RWlpKaWlpeTn55OTk8OkSZN47rnnUh1mlLVr15Kbm8uSJUsOaf/33nuPoqIi5s+fn+TIOmfDhg388Ic/pLS0lKKiIkpLSzn22GOZMWMG999/PxUVFakOscvdd999GGP4z//8z1a3+fOf/0xpaSnp6ekYYygoKKC0tBSXy0V6ejrDhg3j17/+ddRAgVRavXo1EydOpKSkBJfLxdlnn83777/f4eMsX76ck046iYKCAvLz8xkzZgyLFi2KO9/q66+/jsPhCL+fWy4ff/xx1PYzZ84kKysr7rYulwtjDD//+c8P+XegVLtCM9H39WXcuHHSng8++KDdbfqTe++9VwAB5NVXX011OL0aIIMHD44q+/TTT2Xy5MkCyNy5c1MTWBwvvviiGGPkvvvuO6T9N2/eLA6HQ375y18mObJD99hjjwkgV1xxhezbt09ERHw+n2zevFnGjh0rgLz00kspjrLrjR49WgA54ogjxOfztbntpZdeKoBs3LgxXNbQ0BD+XBg1apTU1NR0ccRte+ihhwSQe++9V3w+n9TU1MiMGTMkMzNT3n777YSPM2fOnPD7sL6+XjwejyxdulTS0tLkggsuiNl+48aNctJJJyV8/EsvvVRuu+22uOvWrl0rgPz9739v9zj6HdW/AFslSTlQypOw7lo02eu4UaNGSVlZmQAyffr0VIfTq8VL9kRE9u3bJ2lpaWK1WmX37t3dH1grGhsbU7p/Mnm9XiksLJTDDz88boKza9cusVgsfT7Z27Jli+Tk5MjQoUMFkFdeeaXN7eMleyHTpk0TQO66664uirZ9u3btkrS0NJkyZUpUeW1treTn58uYMWPE7/e3e5y3335bAJkwYULMumuvvVYA+ctf/hJV3tFk795775Wnnnoq7rof/ehHMmrUqISOo99R/Usykz1txlVxbd68mS+++IJly5ZhsVhYuXIlBw4cSHVYfU5JSQnDhw/H5/Pxv//7v6kOJ6yzs/T3pFn+KyoqqKqqoqSkBIsl9iNv4MCBzJ49myOOOCIF0XWfxx9/nOnTp/PTn/40/PxQTZo0CYC33norKbEdiscff5ympiYuuOCCqPKsrCx+8IMfsH37dt588812j7Ny5UoApkyZErPujDPOAOBPf/pTp2K9/vrrueiii2LKy8vLWbVqFVdddVWnjq9UezTZU3EtWbKEGTNmMGLECCZPnozb7Wbp0qXh9SNGjMDhcGCMweVyccUVV4TXnXbaaeTn52O32xkxYkS43O12c/vttzNy5EgKCgooLCxkypQpvPbaa1GvPWbMGAoKCjDGMG/ePJYtW8a4ceMoLCzEGMPMmTOBQEI6a9YsRowYwYABA8jNzWXixIksX7487jl5PB7mzZvHkCFDKCgoYNiwYdxyyy08/PDDGGPIzc2ltLQUt9sd3ufjjz/mxz/+MSUlJRQUFDB48GCuvvpq9u7dm4xfM/D1lADGGP75z39SWlpKVlYWxhjWrVvH/PnzGTlyJBkZGRhjeOKJJ8L7rl27lsmTJ4f7GR133HHceeedeDyemNfZt28fv/jFLxgyZAgul4tBgwZx0kkn8Zvf/IaqqioArrjiinAfw5NPPjlqf7fbzYIFCzjmmGMYMGAARxxxBCeeeCL33HMPBw8eBGDBggWUlpZitVrjTlDe3NzMXXfdxXHHHUdJSQklJSWcdtpprF27Nmq7q666KiqOt99+m5NPPpni4mIGDx7M9ddfT1NTU0K/36KiIux2O++99x7/8z//E3ebe+65h7KyspjyjRs3cvrpp1NSUkJpaSkjRozgvPPO44UXXojZ9vnnn+c73/kOxcXFlJSUMHbsWP7whz9E9XcN9YcLvXcqKiqYOXMmQ4cOxWq1Yozhs88+AwJ/Fw8++CBlZWXh6ztp0iSef/75hM47UlNTE8uWLeMnP/kJl1xyCRaLhdWrV7N///4OHysUG6S283noc2PcuHEx60JlLT9b4tm3bx8AxcXFMetKS0sB2LRpU6ApLMmWLFmCw+Hg4osvTvqxlYqSrCrCnr5oM27iGhoaJDc3VzZv3iwiIkuXLhVAxo4dG7XdddddJ4CsXr065hi/+tWv5Oqrrw4/b25ulpNPPlkKCwtl/fr14vf7paamRq644goxxsjSpUuj9t+4caMAUlZWJj//+c+ltrZWvF6vnH/++XLppZeKSKCZafjw4eHr1tDQIPPmzRNAFi9eHBPTjBkzxBgjDz/8sPh8PmloaJDbb79dSktLBZAlS5ZEbb9161bJycmRKVOmyN69e0VEZPv27TJ8+HAZOHCglJeXJ/w7pZVm3K+++kqcTqfY7Xb56quvwuW33XabAHL88cfL448/Ls3NzXLgwAE56qijwnEuXrxYjDFy4403SkNDg3i9Xlm5cqVkZmbKGWecEdWEtWvXLjn88MNl9OjR8vHHH4tIoLnrmmuuiXvuQEwz1eWXXy4DBgyQHTt2iIiIx+OR+++/P25T3+DBg2PON/Q3kJeXJ+vXrxeRQHPvzTffLMYYeeihh+L+3o488kiZMWOGVFVVid/vl0ceeUQAmTNnTmu/7hiXXHKJAOJwOGTmzJmyfv16aW5ubnOfRx99VIwxMnv2bKmvrxcRkR07dsioUaMk8NH5tZtvvlkAWbRokTQ3N4vf75cXXnhBnE6nzJgxI+bYJ510kgByxhlnyKZNm8Tv98u2bdvEarXKp59+KiIiF154oTidTnnuuefE6/VKY2Nj+O/7N7/5TcLnLiLyzDPPyLHHHht+PmXKFAHk/vvvb3Wftppxp06dKoDcc8897b72smXLxOVyJbxce+21CZ1TcXGxAFJVVRWz7oUXXki4+0no2t16660x69atWxfut1xRUREu37hxoxxzzDHy05/+VEaMGCHFxcUyYsQIufrqq8PXrz1+v1+OPvpoufzyyxPaXkS/o/obtM9e9yd7gxb1/CVZnnrqqag+JI2NjZKXlyeAvPvuu+HyDz/8UAA588wzo/ZvbGyUgoKCqG3vueceAeSBBx6I2ra5uVmOOOIIKSkpkaampnB5KNk76qijxOv1hst37NgRThRuvfVWWbVqVUz848aNkwEDBkSVvfbaawLIj370o5jtJ02aFJPw+P1+Oe644yQ9PT3coT9kzZo1AiT8pSQSm+z5fD55//335cQTT4zb9ymU7M2cOTOq/NVXX5UPP/xQvvjiC3E4HDJ+/PiY1/r1r38tgKxYsSJcdtZZZwkgW7dujdrW5/NFJZCR8bZM9vLz8+Wcc86Jeb0zzzwz/I9BSLxkL/Q3sHDhwphjjBgxQux2u3z++ecxcdhstqhEWERkwIABMnTo0JjjtKa6ulqmT58uxpjwl3dOTo6cd9558pe//CWmL9+XX34pTqdTRo8eHdPv64033ohK9kJ9viZPnhzzurNmzRJAnn322ajyULL3xBNPRJU//vjjUlNTI8uXLxdArr/++phjnnDCCeJwOML/gCRi8uTJUYnZs88+G/5nqjXxkr2qqipZuHBh+J+/UBKcCna7XYCoz42QV155RYCY/nzxvPXWWwJIvO+IG264Ifz3EpnEbdy4UbKysuSxxx6T+vp6cbvdsmHDBjnyyCMlJydHtmzZ0u7rrl+/Pu57si2a7PUvyUz2tBlXxXj88ce57LLLws/T0tLC/WIee+yxcPnIkSM56aSTeOWVV9i1a1e4fPny5QwfPpwxY8aEy55++mkgMBVJJLvdzje+8Q3Ky8vZsmVLTCynnnoqVqs1/HzUqFFMnjwZCDQZTp06NWafY489lj179lBeXh4uW7FiBQCnn356zPbx+uq8++677Nixg+OPP56SkpKodd/85jcBePnll2P2a8vu3bvD0y0UFxczefJk0tPTeemll7jpppvi7vPd7343JtaRI0eyfPlympubw32K2orvwIEDrFmzhgEDBsQ0eVksFl544YWY14mntLSUv/71rzzwwANUV1eHy19++WVOOOGEdvcP/Q2cc845MevOPvtsPB5P3Cb4o446KuauAYMHD+aLL75o9zVDcnJyWLZsGR988AFz586lrKyMmpoali9fzjnnnMOECROi/l6WL1+O2+3mrLPOimmqnDhxIq+88kpC5xUqe+aZZ+LG1fL3ftlll5Gdnd3q+wUC17e5uTmm6bs1n3/+OW+++WZUU+E555xDfn4+27Zt4x//+Eeb+0+bNi08RcjRRx/NCy+8wPz583nzzTfJyMhIKIaebOLEiVx++eW88847XH/99VRVVdHY2MhTTz3Fk08+GT7HzMzM8D4TJkxg165d/OQnPyEjIwOHw8Epp5zC0qVLqampifr8bM3DDz/M8ccfH7cZWqlk00mVVZTPPvuMTZs2sWzZsqjyyy67jD/+8Y8888wz3HPPPTgcDgCuvPJK3njjDR555BEWLlwIBDozX3nllVH7h+admjBhQsxrut1uMjMz+eqrr2LWhfrMxLN//34efPBBXnrpJb788stwP6JQItLQ0BDe9p///CcAhx12WMxx4pWF4t2yZUvcGDIzMzvc32ngwIHh/liJau38Q/EtWrQopvO4z+cjMzMzPG/czp078fv9DBw4MO6x4vVVi+epp57ixz/+Mddeey033HADJ598Mueffz4zZsyI+iJszSeffALA4YcfHrMuVNZyfjIgJtkGcDqdcfsltmfkyJHMnz+f+fPnhwcg/fa3v2XLli1cf/31PPXUU1FxxPudWa3WqH8aDvW8oP3re/7550f9swOBfo+ZmZkJ9xt94oknOP3003G5XOEyp9PJBRdcwEMPPcSSJUv4xje+0er+K1asiOm/2RPk5eVRUVFBfX09Tqczal3o/rH5+fkJHSuUeC1ZsoQRI0aQkZHBt771LV5//XVOPvlkfD5f1LGcTmfMa0IgER84cCDvv/8+n3zyCcOHD4/7ehUVFaxcuZI//vGPiZ6uUp2iyV6CPv9lqiPoHkuWLMFisUTVyoXYbDaqqqpYvXo1P/rRjwA499xzKS4u5rHHHuO2227jo48+4oMPPmD69OlR+4ZqRz766CPy8vISjife6EmAxsZGvv3tb/PFF1+wdOlSTj/99HACOnPmTJ588sm4+0UOvmhLKN7vfe974dF6qdDa+Yfimz9/Ptddd11Cx0r03Fszbtw4PvroIzZs2MCLL77I888/z9q1a1m4cCFr165t9Yuts1r7HXTWEUccwQ033MD3v/99Ro8ezauvvhqzTWd/Z+1p7/quW7cu4WQ8HhHhiSeeoLq6Oiax9Hq9ACxdupR77703/P5Jtueee45f/jLxD9Dp06fz+9//vt3tRo4cSUVFBV9++SUFBQVR67788kuAqAFibTHGMGvWLGbNmhVV7na7qaqqYuzYsdhsiX1dHnbYYezevZu9e/e2+p544oknyMjIYMaMGQkdU6nO0mZcFeb3+3nyySd5+eWX2bt3b8xy3333AdFTNjgcDmbOnMmePXtYtWoVixcv5uKLLyY9PT3q2KEP3d27d8e8bl1dHevXr6e2tjbhWNevX8/HH3/MRRddxNSpU9v9ojr66KMB4tYexitrK14I1JZt3rw54XiTrb34tmzZEq5xGjZsGBaLJfwF2FJVVVV4NG5bvF4vxhgmT57MQw89xFdffcUtt9zC559/zh133JFwzPGaX0OxjRw5st3jHIq8vDzefffduOtGjRpFYWFhuDYoMo54vzO/388XX3wRHg3cFefV1vX1+XysX78+oTt+bNiwAb/fT1VVVcz7ubKyknHjxlFVVdWl/9BMnz497udJa0siiR4Q7s7xzjvvxKwLlYW2OVRvv/02fr8//M9tyKJFi9i5c2fcffbs2QPEr5EOeeSRR7jkkkv6RDO46h002VNhr732Gn6/n1NOOSXu+h//+Mc4HA7Wrl0b9SU4a9YsjDEsWrSIp59+OqYJFwjPMRWvT9Zjjz3GhRdeGLdZpDV2ux2IP/XDp59+GlM2bdo0AF566aWYdevXr48pGzNmDKNHj2bbtm3hJuAQEWH69Ok8++yzCcebbOeddx5Op5MVK1bE3M6pvLycE088kQ8++AAINGWdddZZVFZWxsyL5vV6GT16dFRfzNbY7fbwNBUQSPRvvPFGgITmYAz9Dbz44osx61auXInD4eC8885r9ziHorq6utVpOPbs2UNlZSXf+ta3wmXnnXceaWlprFy5MjCSLcIrr7zCkCFDws34F154IRD/vP7yl79EbZOott4vL7/8MmeeeWZMXPE8/vjjXHzxxa1OkRKaxqgzc+6lyk9+8hPS0tJi3od1dXW89NJLjB49OjwfYEh9fT2VlZUxx/re974X9+/joYcewuVyxXymLVq0KG6C/I9//IPdu3dz9NFHt1qruHHjRnbu3Klz66nulayRHj190alX2nfBBRfILbfc0uY2oZnz77zzzqjy0047TQD5zne+E3e/0LQbWVlZsmrVKvH7/eLz+WTlypWSlZUVM7t8aDRua7cYqq2tlaFDh0pmZqb87W9/E5HAVCC///3v446eE/l66pVHHnlEfD6fNDY2yp133hmeSqO1qVcmTZoUPtbBgwflqquukkGDBsmXX37Z5u8qEq1MvdKa0GjceNNehPzpT38SY4xcffXVUl1dLSIi//rXv2TSpEly+umnR41iDk29UlZWJjt37gyfy8yZM2XYsGGyf//+mHhbjsYF5IILLpDKykoREWlqagrH+fzzz0dt29bUKzk5ObJu3brwMdqbeiXenQpCo1kTBUhmZqY88sgjUltbKyJf3y7thBNOkKysrJjba4WmXpkzZ440NDSIiMg//vEPGTp0aMwo2VtuuUUA+a//+q8OTb3SlgsvvFBsNps8+uij4vF4RETk73//u5SWlsodd9zR7jkfPHhQ0tPTw9c7nsrKSnE4HGKxWGLu4NLW1Cs9Reh2ab/73e/E5/NJbW2tzJgxQzIyMmKuZ11dnbhcLrHb7XFHj48dO1Y+++wzEQmM3r7lllskKysr7m3MBg8eLLm5ubJ69Wrxer3i9/tly5YtMnLkSElLS2vzdzZjxowO3X0jUn//jupv0KlXNNlLpq+++kpcLpdYrVbJy8uTSZMmxWzj9XrF5XJJVlaWAOJ0OsXlcoXnmgtNFfH000+3+jpNTU1yxx13yLHHHisFBQUycOBAOfXUU2Nu2zR58mTJz88Pf0G7XK6484r961//khkzZsiAAQMkNzdXhg0bJrNnz5ZzzjlHACkqKoqaw6q5uVluu+02GTRokOTn58vIkSPlt7/9rTz66KMCyJNPPhnzGp988olcdNFFUlpaKi6XS4488ki58sorZdeuXQn9bs844wxxuVwCiMViCc8lFkrOWqqurhaXyyWZmZkCSH5+vrhcrlbvm7lu3Tr57ne/KwUFBXLYYYfJscceK7fffnvcKTH27NkjP//5z2XQoEHicrlk8ODBcvnll8sXX3wR3ubyyy8Px2u328XlcskzzzwjIiJPP/20nH322TJ06FApLS2Vww47TCZPnix//etfw/vPnz9fXC6XWCyW8PlGXrvIv4Hi4mIpKiqSU089NZywh9x6660xcWzatEk2bdoU/sIGxOVyyfz589u9Dq+++qrccMMNcsIJJ4Sn+snOzpZhw4bJlVde2WpC9Nprr8mUKVOkuLhYXC6XjB49Wh544IG4t+FatmyZTJw4UYqKiqSoqEjKysrkgQceiJrW5ZVXXomJ3+VyRSXmIX6/Xx566CEZO3as5Ofny+GHHy4TJkwIX4+23HfffVJQUCDGGHG5XDFTHokE5t4Lve8BycvLkwsuuECefPJJcblckpaWFvU3+PDDD7f7uqmwcuVKmTBhghQXF0txcbFMnTpV3nvvvZjt3G63jB8/XkpLS+XDDz+MWrdgwQL55je/KYWFhVJSUiJDhw6Vyy+/vNU589555x254YYbpKysTEpLSyU3N1cGDhwol156aZvfIxUVFeJ0OmOm4klUf/2O6q+SmeyZwPH6vvHjx8vWrVvb3ObDDz/kmGOO6aaIVE9x1113cfPNN7NmzRq+//3vpzocpZSKS7+j+hdjzDsiMj4Zx9I+e6rfuOSSS4iX8L/11lukp6dH9dlSSiml+gpN9lS/8cknn/CrX/0qPMKxsbGR3/3ud6xZs4YFCxYkPCeXUkop1ZvoPHuq35gzZw5//vOfmTRpEm63m4aGBsaMGcPzzz/fZaNAlVJKqVTTZE/1G+eeey7nnntuqsNQSimlupU24yqllFJK9WGa7CmllFJK9WGa7LXQX6aiUUop1Xvod5PqDE32IlitVjweT6rDUEoppaJ4vV5sNu1mrw6NJnsRsrOzqampSXUYSimlVJTa2lrS0tJSHYbqpTTZi1BQUMCBAweorKykublZq82VUkqllIjQ0NBAZWUlxcXFqQ5H9VJaJxzB6XQyaNAg9u/fz2effYbP50t1SEoppfo5p9OJy+XSmj11yDTZa8HpdDJgwAAGDBiQ6lCUUkoppTpNm3GVUkoppfowTfaUUkoppfowTfaUUkoppfqwbk32jDElxphnjDEfB5cXjDFHJLiv3Riz0BjzkTFmhzHmv40xk7o6ZqWUUkqp3qzbkj1jjANYBziAUcCxQD2w0RiTlcAhHgCmA98RkeOAx4G1xpiyLgpZKaWUUqrX686avUuBMcCNIuIVER9wI3AkcHVbOxpjRgCzgN+ISAWAiDwKfArc0aVRK6WUUkr1Yt2Z7J0L7BKRf4cKRGQv8EFwXVvOAQywsUX5BmBKgjWDSimllFL9Tncme2MI1MS19CkwOoF9/cCuOPvaCDQJK6WUUkqpFrpzUuUi4J045TVAhjEmXUQa29i3Idj023JfgMJ4OxljZhFo/gVwG2N2dDBm1TMUAZWpDkIdMr1+vZtev95Lr13vNiJZB+rTd9AQkYeBhwGMMVtFZHyKQ1KHQK9d76bXr3fT69d76bXr3YwxW5N1rO5sxq0EsuOU5xCotWutVi+0b4YxxhpnX4CqJMSnlFJKKdXndGeytx0YEqd8KPBeAvtagIFx9vUSGOShlFJKKaVa6M5kbwUw2BgzJFRgjHEBxwAvRm5ojHEZYyJj+wsgwMktjnkKsFZE6hJ4/Yc7HrLqIfTa9W56/Xo3vX69l1673i1p18+ISLKO1fYLBSZV3gp8CFxIYHTtY8Ak4BuhhM0YMxH4O/CwiFwdsf9iAsndRBGpNMZcBjwETBCRbd1yEkoppZRSvUy31eyJSDPwXcBHoNn1QwJ97k5tUTNXB1QDe1oc4hpgObApOKp2FjBFEz2llFJKqdZ1W82eUkoppZTqft3ZZ08ppaIYY243xogxZmaqY1FKqe5gjBlgjPmbMabbatt6dbJnjCkxxjxjjPk4uLxgjDkiwX3txpiFxpiPjDE7jDH/bYyZ1NUxq68d6vULvlHmG2O2B6/dR8aYFcaY9u7EopKkM++9iGMcAczuohBVGzp7/Ywx/2GMWWWM+b/g++9jY8xvuzJmFdDJ770BxphHg9dsuzHmfWPMzcYYe1fHrQKMMdOA/wGOOsT9rzPGfBC8fv9njDk7kf16bbIXHPCxDnAAowjcMq0e2JjgvXIfAKYD3xGR44DHgbXGmLIuCllF6OT1uw24ADgzeO3KCPQF3awJX9dLwnsv5E4C97dW3aiz188Y823gFeC3IjJWREYC9wPnd13UCjp37YIzXKwBvkVgoOMY4CLgVuCuroxbRbmRwPiFTR3d0RhzE/CfwA+C1+9GYLkx5vvt7iwivXIBriAwHcuREWWlBL7057Sz7wgCo4F/0qL8feCvqT63/rB08votBi5vUXZU8HgPpPrc+vrSmWsXsf044F/A94LHmpnq8+ovSyffe4bA4Lo5LcrtwPdTfW59fenktTs2uO+vWpSvAvak+tz6ywLYgo9PBFKwhPfLI5DYL2hR/lfg/fb277U1e8C5wC4R+XeoQET2Ehjpe247+55D4ENrY4vyDcCUDtZOqEPTmev3CwI1sZG+Cj7mJy1C1ZrOXLuQ3wG3AO7kh6fa0ZnrNwkYCbwcWSgiHhF5JdmBqhiduXbe4GPL26TagJZ3p1JdRES87W8V1+lABvHzlmONMSPb2rk3J3tjgE/jlH8KtNeUN4ZAzd6uOPvaCPwHpLrWIV8/EfGKiL9F8fDg4+udD021ozPvPYJ9TNKB55Icl0pMZ67ft4OPucE+e+8H+w7dboxJT2qUKp7OfG5+AiwFrgzd3MAYcyqBJsUHkhql6gpjgo8tr/+nLdbH1ZuTvSKgNk55DYH76Lb1wVNE4H68vjj7AhQmIT7Vts5cv3hmEWiGf6qzgal2HfK1C3YEvxu4XpQcjwAAAAaFSURBVIJtEKrbdea9F7pl5bPAHSIyCrgYmEmgOVB1rc5+bl5KoN/eTmPMV8BK4DoRWZjcMFUXKAo+trz+CeUtvTnZUwoAY8xkAoNtzhcRbRbs2a4m0L/krVQHog5JWvDxMRHZAiAi7xJI4L9rjDkpZZGpNhlj0gg0AZ4ADBGRwwjcgvT/GWNuSWVsquv15mSvEsiOU55DoNausZ19M4wxLfsp5AQfq5IQn2pbZ65fmDHmP4Angaki8kES41OtO6RrZ4zJA/4fgRFkKnU6894L1Sq0vHPRP4KPx3cyNtW2zly7nxDoczlHRL4EEJH/A+4FFupMFD1eZfCx5fVPKG/pzcnedmBInPKhwHsJ7Gvh6yaJyH29BDq7qq7VmesHgDFmDIFmiBki8t/JC02141Cv3bcIvL+WG2O2GWO2AY8G1y0Ils1NaqQqns689z4KPrb87vC1Uq6SqzPXLtSnb2eL8k8IDFjURL1n2x58HNKifGiL9XH15jfmCmBwqKMpgDHGBRwDvBi5oTHGFZxjKOQvBIagn9zimKcAayX6Xr2qa3Tm+oUSvVXAxaEmweCEoX/q4rjVIV47EfmbiAwUkbLQAlwe3HRusGxBt5xB/9aZ994aAoldy87gxwUf3052sCpKZ65defBxUItjDg4+aotWD2KMKQzOqxjyN6CB+HnLByLyEW1J9ZwznZirxkEgk32OwAhaC7CEwH8tWRHbTSTw4fTHFvsvBj4GioLPLwMagbJUn1t/WDpz/Qj8h1oRvIYXRSzXAa+n+tz6+tLZ916LY52MzrPXq64f8F/AHmBY8PnhwX3Xpvrc+vrSyc/NoQQ6868FsoNlg4B/EpjzMj3V59efFtqYZy94rZqAV1qU3xT87jsy+Pw0wEMCc1y2nG+n1xCRZmPMd4H7CDS7CrADOFWia+bqgGoCH06RriFwJ4ZNxhgPgb4oU0SkZV8U1QU6ef3mExiZdGVwifRGlwWtgKS89zDGlBD40gnNabnAGHMdgcmyt3Zl/P1dEq7fHAL9h9YYY3wEJlR+kcDnqepCnbl2IvKpMeYEYB7wtjGmmcC1exVYKAn2k1adY4y5h8B0N4OCz0M5xwki0hz8uRHYz9fzxwIgIr8xxjQBLxtjvAQS+vMkgTkuTTA7VEoppZRSfVBv7rOnlFJKKaXaocmeUkoppVQfpsmeUkoppVQfpsmeUkoppVQfpsmeUkoppVQfpsmeUkoppVQfpsmeUkoppVQfpsmeUqrPM8Ycbox53RhzMLi8Hlz+zxjzgTHm5ymIaZUxZq8x5vXg8/SIGJ/o7niUUn1Xr72DhlJKJUpEvgRODiVWInJyaJ0xZhrwojGmWkSe7saYfhhM6oYEnzdGxqiUUsmiNXtKqX5NRFYAB4EfpjoWpZTqClqzp5RSgc9Cf+iJMebbwF2Ak8CNxj8HbhCRvRHbnAHMBUywaD+wSEReDa6/BziZwH1Ks4C3gP8UkfquPhmllIqkNXtKqX7NGPMrwAH8Mfh8DPAacL+IfAs4ETgArDPG2ILbTAFWE0jevgl8C/gncHXEoS8HfigipwCTgOOAu7vlpJRSKoIme0qp/qYsOBBikzGmBjgHGCMirwfX/xr4VEReBBARAR4ikKxNCW5zK7BZRNZHbPNbYEPE64wVka+C693AC8HXUkqpbqXNuEqp/mZbaICGMeZEYD1wNl/Xuo0FXC0GSlgINOW6gs/HAc9HHlREdgP3RxSNNcb8kUATrhcoBQ5L5okopVQiNNlTSvVbIvJ3Y8zDwH8aYx4VkargqvcjR+x2lDHmbAI1eReJyDPBspnAkk6GrJRSHabNuEqp/u43BPrsXRd8/g4w3BhjjdzIGHOPMea4yG1arB9kjLk2+HQy4AOejdjEkezAlVIqEZrsKaX6NRH5gkCN2y+MMTkEmnNzgF+FtjHGTAXOBD4KFi0ATjDGnBZcbwHmA4XB9e8CViC03g6c2+Uno5RScZhAv2KllOq7jDGHA88AZcGibQRG0r4VXD8E+AT4GHgZWEmgxq8YKAcqCUy9sivimGcSGKhhIVCL93rwmL5g8ncncCHwb6CKwNQsPwXeAK4kMKDjm0BaMJ4fAquCMTYB/y0i05L9u1BK9T+a7CmllFJK9WHajKuUUkop1YdpsqeUUkop1YdpsqeUUkop1YdpsqeUUkop1YdpsqeUUkop1YdpsqeUUkop1YdpsqeUUkop1YdpsqeUUkop1Yf9f8wnExcZsTBGAAAAAElFTkSuQmCC\n","text/plain":["<Figure size 720x468 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]}]}