fullyConnected.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importer les librairies de machine learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import poutyne as pt\n",
    "from poutyne import Model\n",
    "import torch\n",
    "from torch import nn\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "\n",
    "from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score, confusion_matrix, ConfusionMatrixDisplay\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ajuster les hyperparamètres"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_ratio = 0.2\n",
    "valid_ratio = 0.5\n",
    "\n",
    "hidden_size = 512\n",
    "output_size = 8\n",
    "dropout = 0.0\n",
    "\n",
    "n_epochs = 30\n",
    "batch_size = 32\n",
    "learning_rate = 0.1\n",
    "min_lr=0.00001\n",
    "momentum = 0.9\n",
    "\n",
    "gamma = 0.1\n",
    "patience = 1\n",
    "\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Charger le dataframe traité, le séparer en datasets et créer les DataLoaders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load data/dataframe.csv file as a pandas dataframe\n",
    "df = pd.read_csv('data/dataframe.csv')\n",
    "\n",
    "input_size = len(df.columns) - 2 # Retirer le ID et l'étiquette\n",
    "\n",
    "# Convert into a tensor\n",
    "dataset = torch.tensor(df.values)\n",
    "\n",
    "# Split into train and test\n",
    "train_data, test_data = train_test_split(dataset, test_size=test_ratio)\n",
    "\n",
    "# Split train into train and validation\n",
    "test_data, val_data = train_test_split(test_data, test_size=valid_ratio)\n",
    "\n",
    "# Split train into X and y without the firs column\n",
    "X_train, y_train = train_data[:, 1:-1].float(), train_data[:, -1].long()\n",
    "X_val, y_val = val_data[:, 1:-1].float(), val_data[:, -1].long()\n",
    "X_test, y_test = test_data[:, 1:-1].float(), test_data[:, -1].long()\n",
    "\n",
    "train_dataset = TensorDataset(X_train, y_train)\n",
    "val_dataset = TensorDataset(X_val, y_val)\n",
    "test_dataset = TensorDataset(X_test, y_test)\n",
    "\n",
    "\n",
    "train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
    "val_loader = DataLoader(val_dataset, batch_size=batch_size, shuffle=True)\n",
    "test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(74992, 32)\n"
     ]
    }
   ],
   "source": [
    "# print shape of df\n",
    "print(df.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Créer l'algorithme FC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class FCNetwork(nn.Module):\n",
    "    def __init__(self, input_size, output_size):\n",
    "        super(FCNetwork, self).__init__()\n",
    "        self.dropout1 = nn.Dropout(dropout)\n",
    "        self.fc1 = nn.Linear(input_size, hidden_size)\n",
    "        self.bn1 = nn.BatchNorm1d(hidden_size)\n",
    "\n",
    "        hidden_size_2 = hidden_size\n",
    "        self.dropout2 = nn.Dropout(dropout)\n",
    "        self.fc2 = nn.Linear(hidden_size, hidden_size_2)\n",
    "        self.bn2 = nn.BatchNorm1d(hidden_size_2)\n",
    "\n",
    "        hidden_size_3 = hidden_size // 2\n",
    "        self.dropout3 = nn.Dropout(dropout)\n",
    "        self.fc3 = nn.Linear(hidden_size_2, hidden_size_3)\n",
    "        self.bn3 = nn.BatchNorm1d(hidden_size_3)\n",
    "\n",
    "        hidden_size_4 = hidden_size // 4\n",
    "        self.dropout4 = nn.Dropout(dropout)\n",
    "        self.fc4 = nn.Linear(hidden_size_3, hidden_size_4)\n",
    "        self.bn4 = nn.BatchNorm1d(hidden_size_4)\n",
    "\n",
    "        hidden_size_5 = hidden_size // 8\n",
    "        self.dropout5 = nn.Dropout(dropout)\n",
    "        self.fc5 = nn.Linear(hidden_size_4, hidden_size_5)\n",
    "        self.bn5 = nn.BatchNorm1d(hidden_size_5)\n",
    "\n",
    "        self.fc6 = nn.Linear(hidden_size_5, output_size)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.dropout1(x)\n",
    "        x = F.relu(self.bn1(self.fc1(x)))\n",
    "        x = self.dropout2(x)\n",
    "        x = F.relu(self.bn2(self.fc2(x)))\n",
    "        x = self.dropout3(x)\n",
    "        x = F.relu(self.bn3(self.fc3(x)))\n",
    "        x = self.dropout4(x)\n",
    "        x = F.relu(self.bn4(self.fc4(x)))\n",
    "        x = self.dropout5(x)\n",
    "        x = F.relu(self.bn5(self.fc5(x)))\n",
    "        x = self.fc6(x)\n",
    "        y = F.log_softmax(x, dim=1)\n",
    "        return y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Créer le modèle Poutyne"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "network = FCNetwork(input_size, output_size)\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.SGD(network.parameters(), lr=learning_rate, momentum=momentum)\n",
    "scheduler = pt.ReduceLROnPlateau(monitor=\"val_acc\", mode='max', patience=patience, factor=gamma, min_lr=min_lr, verbose=True)\n",
    "model = Model(network, optimizer, criterion, device=device, batch_metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Entraîner le modèle sur les données d'entraînement"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[35mEpoch: \u001b[36m 1/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.72s \u001b[35mloss:\u001b[94m 1.497525\u001b[35m acc:\u001b[94m 45.828680\u001b[35m val_loss:\u001b[94m 1.306995\u001b[35m val_acc:\u001b[94m 51.693333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 2/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.17s \u001b[35mloss:\u001b[94m 1.284816\u001b[35m acc:\u001b[94m 53.372894\u001b[35m val_loss:\u001b[94m 1.147441\u001b[35m val_acc:\u001b[94m 56.880000\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 3/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.55s \u001b[35mloss:\u001b[94m 1.169417\u001b[35m acc:\u001b[94m 57.345024\u001b[35m val_loss:\u001b[94m 1.039714\u001b[35m val_acc:\u001b[94m 62.026667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 4/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.31s \u001b[35mloss:\u001b[94m 1.093895\u001b[35m acc:\u001b[94m 60.128682\u001b[35m val_loss:\u001b[94m 1.019630\u001b[35m val_acc:\u001b[94m 62.546667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 5/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.34s \u001b[35mloss:\u001b[94m 1.042589\u001b[35m acc:\u001b[94m 61.695531\u001b[35m val_loss:\u001b[94m 0.941383\u001b[35m val_acc:\u001b[94m 65.746667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 6/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.50s \u001b[35mloss:\u001b[94m 0.993607\u001b[35m acc:\u001b[94m 63.554081\u001b[35m val_loss:\u001b[94m 0.944207\u001b[35m val_acc:\u001b[94m 65.520000\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 7/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.42s \u001b[35mloss:\u001b[94m 0.959809\u001b[35m acc:\u001b[94m 64.689214\u001b[35m val_loss:\u001b[94m 0.885580\u001b[35m val_acc:\u001b[94m 67.133333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 8/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.39s \u001b[35mloss:\u001b[94m 0.923294\u001b[35m acc:\u001b[94m 66.041038\u001b[35m val_loss:\u001b[94m 0.850452\u001b[35m val_acc:\u001b[94m 69.013333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m 9/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.67s \u001b[35mloss:\u001b[94m 0.896420\u001b[35m acc:\u001b[94m 67.041155\u001b[35m val_loss:\u001b[94m 0.897195\u001b[35m val_acc:\u001b[94m 66.773333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m10/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.63s \u001b[35mloss:\u001b[94m 0.871822\u001b[35m acc:\u001b[94m 67.931259\u001b[35m val_loss:\u001b[94m 0.835488\u001b[35m val_acc:\u001b[94m 68.826667\u001b[0m\n",
      "Epoch    10: reducing learning rate of group 0 to 1.0000e-02.\n",
      "\u001b[35mEpoch: \u001b[36m11/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.52s \u001b[35mloss:\u001b[94m 0.728470\u001b[35m acc:\u001b[94m 73.068525\u001b[35m val_loss:\u001b[94m 0.709396\u001b[35m val_acc:\u001b[94m 73.933333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m12/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.65s \u001b[35mloss:\u001b[94m 0.683783\u001b[35m acc:\u001b[94m 74.433684\u001b[35m val_loss:\u001b[94m 0.704485\u001b[35m val_acc:\u001b[94m 74.093333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m13/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.60s \u001b[35mloss:\u001b[94m 0.661561\u001b[35m acc:\u001b[94m 75.310453\u001b[35m val_loss:\u001b[94m 0.680322\u001b[35m val_acc:\u001b[94m 75.013333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m14/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.39s \u001b[35mloss:\u001b[94m 0.643952\u001b[35m acc:\u001b[94m 76.002200\u001b[35m val_loss:\u001b[94m 0.682329\u001b[35m val_acc:\u001b[94m 75.293333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m15/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.61s \u001b[35mloss:\u001b[94m 0.632517\u001b[35m acc:\u001b[94m 76.375577\u001b[35m val_loss:\u001b[94m 0.654020\u001b[35m val_acc:\u001b[94m 76.146667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m16/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.89s \u001b[35mloss:\u001b[94m 0.617780\u001b[35m acc:\u001b[94m 77.043988\u001b[35m val_loss:\u001b[94m 0.654673\u001b[35m val_acc:\u001b[94m 75.933333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m17/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.41s \u001b[35mloss:\u001b[94m 0.605191\u001b[35m acc:\u001b[94m 77.350691\u001b[35m val_loss:\u001b[94m 0.638548\u001b[35m val_acc:\u001b[94m 76.826667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m18/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.88s \u001b[35mloss:\u001b[94m 0.597757\u001b[35m acc:\u001b[94m 77.700732\u001b[35m val_loss:\u001b[94m 0.638598\u001b[35m val_acc:\u001b[94m 77.026667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m19/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.36s \u001b[35mloss:\u001b[94m 0.585041\u001b[35m acc:\u001b[94m 78.202457\u001b[35m val_loss:\u001b[94m 0.637489\u001b[35m val_acc:\u001b[94m 77.333333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m20/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.33s \u001b[35mloss:\u001b[94m 0.578553\u001b[35m acc:\u001b[94m 78.340806\u001b[35m val_loss:\u001b[94m 0.626575\u001b[35m val_acc:\u001b[94m 77.733333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m21/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.31s \u001b[35mloss:\u001b[94m 0.570083\u001b[35m acc:\u001b[94m 78.805861\u001b[35m val_loss:\u001b[94m 0.634494\u001b[35m val_acc:\u001b[94m 77.226667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m22/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.31s \u001b[35mloss:\u001b[94m 0.554726\u001b[35m acc:\u001b[94m 79.450936\u001b[35m val_loss:\u001b[94m 0.618259\u001b[35m val_acc:\u001b[94m 78.253333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m23/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.32s \u001b[35mloss:\u001b[94m 0.553750\u001b[35m acc:\u001b[94m 79.315920\u001b[35m val_loss:\u001b[94m 0.614308\u001b[35m val_acc:\u001b[94m 78.506667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m24/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.75s \u001b[35mloss:\u001b[94m 0.542881\u001b[35m acc:\u001b[94m 79.799310\u001b[35m val_loss:\u001b[94m 0.614470\u001b[35m val_acc:\u001b[94m 78.520000\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m25/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.42s \u001b[35mloss:\u001b[94m 0.535825\u001b[35m acc:\u001b[94m 80.089344\u001b[35m val_loss:\u001b[94m 0.609613\u001b[35m val_acc:\u001b[94m 78.720000\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m26/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m8.13s \u001b[35mloss:\u001b[94m 0.525276\u001b[35m acc:\u001b[94m 80.371043\u001b[35m val_loss:\u001b[94m 0.599730\u001b[35m val_acc:\u001b[94m 79.013333\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m27/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.70s \u001b[35mloss:\u001b[94m 0.518412\u001b[35m acc:\u001b[94m 80.827763\u001b[35m val_loss:\u001b[94m 0.588596\u001b[35m val_acc:\u001b[94m 79.986667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m28/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.42s \u001b[35mloss:\u001b[94m 0.512438\u001b[35m acc:\u001b[94m 80.802760\u001b[35m val_loss:\u001b[94m 0.592086\u001b[35m val_acc:\u001b[94m 79.386667\u001b[0m\n",
      "\u001b[35mEpoch: \u001b[36m29/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.38s \u001b[35mloss:\u001b[94m 0.507493\u001b[35m acc:\u001b[94m 81.254480\u001b[35m val_loss:\u001b[94m 0.595855\u001b[35m val_acc:\u001b[94m 79.146667\u001b[0m\n",
      "Epoch    29: reducing learning rate of group 0 to 1.0000e-03.\n",
      "\u001b[35mEpoch: \u001b[36m30/30 \u001b[35mTrain steps: \u001b[36m1875 \u001b[35mVal steps: \u001b[36m235 \u001b[32m7.57s \u001b[35mloss:\u001b[94m 0.468361\u001b[35m acc:\u001b[94m 82.661310\u001b[35m val_loss:\u001b[94m 0.593050\u001b[35m val_acc:\u001b[94m 79.986667\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(train_loader, val_loader, epochs=n_epochs, callbacks=[scheduler])\n",
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 864x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss_per_epoch = []\n",
    "acc_per_epoch = []\n",
    "val_loss_per_epoch = []\n",
    "val_acc_per_epoch = []\n",
    "for epoch in history:\n",
    "    loss_per_epoch.append(epoch['loss'])\n",
    "    acc_per_epoch.append(epoch['acc'])\n",
    "    val_loss_per_epoch.append(epoch['val_loss'])\n",
    "    val_acc_per_epoch.append(epoch['val_acc'])\n",
    "\n",
    "plt.figure(figsize=(12, 4))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(loss_per_epoch, label='Erreur d\\'entraînement')\n",
    "plt.plot(val_loss_per_epoch, label='Erreur de validation')\n",
    "plt.title('Erreur par époque')\n",
    "plt.legend()\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(acc_per_epoch, label='Exactitude d\\'entraînement')\n",
    "plt.plot(val_acc_per_epoch, label='Exactitude de validation')\n",
    "plt.title('Exactitude par époque')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Évaluer le modèle sur les données de test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Obtenir les prédictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[35mPrediction steps: \u001b[36m235 \u001b[32m0.37s \u001b[0m                                           \n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict_generator(test_loader, has_ground_truth=True)\n",
    "\n",
    "# Obtenir la classe ayant le plus grand score\n",
    "predicted_classes = predictions.argmax(axis=1)\n",
    "\n",
    "# Obtenir les labels test\n",
    "ground_truth = test_data[:, -1].long().numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculer la matrice de confusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = [\n",
    "    'Another party (please specify)',\n",
    "    'Bloc Qu<e9>b<e9>cois',\n",
    "    'Conservative Party',\n",
    "    \"Don't know/ Prefer not to answer\",\n",
    "    'Green Party',\n",
    "    'Liberal Party',\n",
    "    'ndp',\n",
    "    \"People's Party\"\n",
    "]\n",
    "classes_idx = [0, 1, 2, 3, 4, 5, 6, 7]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAATgAAAEGCAYAAADxD4m3AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8rg+JYAAAACXBIWXMAAAsTAAALEwEAmpwYAABNaklEQVR4nO2dd3hU1daH3zUlvZACJITeQpMm1YqCit3vXnu99oJdVOx6vaLX3gtXLxbsvdFExU6RXqSEFgiBkN7LzKzvjzNAUJJMkpkhmbvf55lnZs7ss39nnzNnnV3XElXFYDAYQhHbgT4Ag8FgCBTGwBkMhpDFGDiDwRCyGANnMBhCFmPgDAZDyOI40AdQm6REm3buFJxD2rA8Jig6BkNLopIyqrVKmpPHcUdFa16+26e0i5ZXzVLV8c3Raw4tysB17uTguxntgqJ1ZsfRQdExGFoS8/XbZueRm+9m/qyOPqV1pm5IbrZgM2hRBs5gMLQGFLd6DvRB+IQxcAaDoVEo4KF1LBAwBs5gMDQaD6YGZzAYQhBFqTFNVIPBEIoo4DZNVIPBEKqYPjiDwRCSKOBuJV6IjIEzGAyNpnX0wLUCAzf91RS+fbc9qjD23J2ceNkOPniiI9++0564pBoAzrk9k6FjC8lYEsMrt3e3dlQ44+ZtjDg+3y/HMWxMMVc9uB27TZnxbiIfPN/eL/karcARqmUL9nn8M4qaPjgAERkPPAPYgVdV9ZHG7J+5JpJv323P5K9W4HB6mHx+X4YeXQjAiZdv55SrsvdJ36lPOY9MX47dAQU7ndx67CAOPiYfezNLabMpEyZnccfZ3cnNdvLc9PXMmxVP5vqI5mVstAKiFWy9UNWqC1WoaR32LXCL7UXEDrwAHA/0A84RkX6NySMrI5JeQ0oIj/Rgd0DfUcUsmJlYZ/rd6QBqqmyI+OcqpA8pZ/vmMHZkhuOqsTH38zaMPq7IL3kbrcAQqmUL9nncP4Lbx9eBJpDeREYAGaq6UVWrgfeAUxuTQaf0Cv6YH0dJgYOqChtLvksgb3sYALNeT2HiuIG8eEsPSgvte/ZZvziGm48exC3jBnH5w5uaXXsDSEqpYZdXFyA320lyak3zMzZaAdEKtl6oatWFAh717XWgCWQTNQ3YWuv7NmDknxOJyBXAFQAd0+z7/NaxVwWnXrOdf53Tl4hoD136lWFzKMdeuJPTb9wGAu8/1ok3H+zKNU9sAKDX0FKe/G4Z29ZH8sKNPRl8VAFhEc0707KfB1GgBpGMVuvTC1Wt+mgJtTNfCGQNbn9n4C+XQlWnqOowVR2WnPTXwzn6nBz+PXMFD3y8ipg2LlK7VdKmbQ02O9hsMPbcHDYs/avro469KoiIcrN1bVSzC5Kb7aRth+o935NTa8jb4Wx2vkYrMFrB1gtVrbqwJvqaJuo2oFOt7x2B7Y3NpCjXqmTmZoWxYEYSh56aS8HOvRd0wcxEOqWXA5CTGY7bZW3ftS2M7RsjadupqqnHv4e1S6NI61ZN+05VOJwexpxayLzZ8c3O12gFRivYeqGqVRcK1KjNp9eBJpBN1IVALxHpBmQBZwPnNjaTJ65Ip6TAgcOhXPrQRmLauHnu+m5sXhWNiNK2UxVXPLIRgDULYvnsxT7YHYrNZqWPS3Q1uyAet/DCXWlMfmcjNjvMfi+RLesCM2pltFqfXqhq1YUiuFuJM3AJZFxUETkBeBprmsh/VfWh+tIPGRSmxuGlwRA45uu3FGt+s9qOfQeG6+tfdvAp7aiumxep6rDm6DWHgM6DU9XpwPRAahgMhuCyuw+uNdDiVzIYDIaWhuBuAf1rvmAMnMFgaBSWR19j4AwGQwiiKlSrveGELQBj4AwGQ6PxmD44g8EQiliDDKaJajAYQhIzyGAwGEIUM8jQRDYsjwnaBNynN/8aFB2AG7seEjQtgyEYuNX0wRkMhhBEEWq0dZiO1nGUBoOhxWAGGQwGQ8iiiGmiGgyG0MUMMhgMhpBEFTNNxGAwhCbWIINZqmUwGEIUM8hgMBhCEkXwtJJBhtZhhg0GQ4vCjc2nV0OIyE0iskpEVorIuyISISKJIvKNiKz3vifUSn+HiGSIyFoROa6h/Ft1DW7YmGKuenA7dpsy491EPni+fbPy++G/qfz2XntQGHX2TsZcms3Sr5OY+XQndmZEctPny+k8sAyALUtjeP+OHtaOCuNv3MrA8fnNLRI3P5nJyHElFOY6uPLo9Gbn1xD+Pod14Qz38MQnGTjDFLtD+enrNrz1eEpAtADadqjm1mcySWjnQj0wfVoSn73WNiBawbxmwf5/7A8rLmrz60YikgZcD/RT1QoR+QArdks/4FtVfUREJgGTgNu9gePPBvoDHYA5ItJbVd11aQQysv1/RSRHRFYGIn+bTZkwOYu7z+vG5WPSOerUQjr3qmxyftlro/jtvfbc/Plybp2xlNXfJbBrUwQp6eVc/PIauo8o3id9ano5t3y5jNtmLOPKN1fzwV099kT0ag6z30/krvO6NT8jH/D3OayPmirhtjN6cPUx6Vx9TDrDxpTQZ2hZQLQA3C5hyj87cPmRfbjhpF6c/I/cgJUtmNcsmFp149fI9g4gUkQcQBRW5L1TgTe8v78BnOb9fCrwnqpWqeomIAMrwHydBLKJ+jowPlCZpw8pZ/vmMHZkhuOqsTH38zaMPq6oyfntzIik65ASwiI92B3QY2Qxy2clktKzgvY9/npj7E4H4KqygfgneM/K+TGUFASnYu3vc1g/QmW5NfLmcCp2pwY0YHF+jpOMFVZM3IoyO1szIgIWAT6Y1yyYWnVhhQ20+/SqNx/VLOBxIBPIBopUdTbQXlWzvWmygd2RqPYXTD6tPo2AGThV/RFofputDpJSati1PWzP99xsZ7P+wCnp5WxYEEdZgYPqChurv0+gMDu83n02L4nhkWMG8+/jBnPmvzbuMXitBX+fw4aw2ZQXv1nL+8tXseTHGNYuiQ6YVm3ad6ymx4AK1ixufhBwg+XR16M2n15Asoj8Xut1xe58vH1rpwLdsJqc0SJyfj3SPgWTr80BvyW9Bb4CIALf/4Cyn6I2p0aQ0rOCsVdl8dL5/QiLdpPWtwybvf4Muw4pZdI3S9mREck7t/Sk75gCnBEBrJb4GX+fw4bweIRrjkknOs7Nfa9tokt6BVvWRgZOEIiIcnPPq5t5+d4OlJe2jrlbrYFGTPTNrSds4Dhgk6ruAhCRT4BDgJ0ikqqq2SKSCuR40zc6mPwBH0VV1SmqOkxVhzmpv8ZUm9xsJ207VO/5npxaQ94OZz17NMyos3KY+PVyrv9gFVFtXLTt5lufTUrPCsIiPWSva101hECcQ18oK7az7LcYhh9VElAdu0O559XNfPdJAr/MaBNQrf8lLH9w4tOrATKBUSISJSICjAX+AL4ALvKmuQj43Pv5C+BsEQn3BpTvBSyoT+CAG7imsnZpFGndqmnfqQqH08OYUwuZNzu+WXmW5Fo3d0FWGMtnJjL0lF11ps3bGr5nUCF/Wzg5GyNJ7FjVLP1gE4hzWBfxiS6i46zBrrAID0MPL2VrRiAjsis3P7GVresj+GRKYEZP/3exPPr68qoPVZ0PfAQsBlZg2aMpwCPAMSKyHjjG+x1VXQV8AKwGZgIT6htBhRbQRG0qHrfwwl1pTH5nIzY7zH4vkS3rmnfDTL06nbICB3aHcvqDm4iKd7N8ZiIf39+N0nwnUy7pS1rfMq5+6w82Lozj25fSsDkUm005/cGNxCQ2fxh10otbGDi6lPhEF9N+X81bT7Rn1rtJzc53fwTiHNZFYvsaJj6Tic0GNhv8+GU88+fEBUQLoP+IMsadUcDG1RG8+M1aAKY+nMrC7/yvGcxrFkyturCmifhnoq+q3gfc96fNVVi1uf2lfwh4yNf8RQPU6SIi7wJjgGRgJ3Cfqr5W3z5xkqgjZb/l8jvGo6/hf5H5+i3Fmt8s65TSP1EvfMe3+/SxwR8tqqcPLuAErAanqucEKm+DwXBgMe6SDAZDSGK5S2oda1GNgTMYDI2mtSy2NwbOYDA0CsubiGmiGgyGEMRaqmUMnMFgCElMDc5gMIQwPqxSaBEYA2cwGBqFGUVtKgLiCM4h3ZR+VFB0ADI/7B00LYBud5QGTyw3YA5j/oK6PUHTkv15Iggg6q53xZHfkHL/NC1NE9VgMIQkrSkmgzFwBoOhUSjgMjU4g8EQqpgmqsFgCE3UNFENBkOIstvhZWvAGDiDwdBoTA3OYDCEJP50eBlojIEzGAyNQhFcHjPIYDAYQhTTBxcAbnpsMyPHFlGY5+CqY/oDcNmd2xg5rhBXjY3tW8J5cmIXyoqbX6zk1CpufWIjCW1rUI8w/d22fP56CpfdkcnIsYW4aoTtWyJ48tZulJU0TU/K3CS9tA3n1ioQyLs6DVeHcJKe2opjVzWutmHk3twZjbHC3Tm3VJD4ShZS4QGBHY/0hLCGn6Q33r6YEYfsoLAgnGv+Ybmajomt5o77F9IutZyc7Cgevm84paVhxMZVc+c/F9C7TwFzZnbmpacHNalstbHZlGc+XEzeznDuv2YAl0zcyMgxebhqbGRvjeCpu9KbfA534wzz8Njby3GGebDb4edZSUx7rgvd+5Ry3QMbcIZ7cLuFF+7vwboVsc0u055yfbSEvJxw7r+qPzHxNdzx5BrapVWSkxXBwzf1obS4+VHKnGEeHntnpVU2h/LzzCSmPdt5z+9/vzSLyyZt4awRwykuCHxUNLT1NFEDVs8UkU4i8r2I/CEiq0Tkhubm+c2HSdx9Ya99ti3+KY4rj+nP1cf1I2tTOGdN2NFcGQA8LuE/D3XmimMGcuPf+nHyhTvp3LOCxT/Hc+VxB3H18QeRtSmCs67JbrJGwtTtVAyJJfuZ3mQ/1pOajhHEfbaLqoOiyX4unaqDoon/zBsS0q0kPbuN/CvS2PFUb3Ie6A523/5kc2Z25p5b940LceZ561i6uC2Xn3sMSxe35Yzz1wNQXW3jrdf68tqLA5pcrj9z6gVZbN2wN6Tikl/bcPWpw5jwfweTtTmSMy/PbLZGTbUw6aKDmHDqUCacNpiDDy+gz6BiLr11M2+/0IlrTxvCtGc6c+mtm5qttZtTL8xi68a95Trz8m0sndeGy8cPZ+m8Npxx+Ta/6NRUC5Mu7M+EUwYz4ZRBHHxEIX0GWyEXk1OqGHJoETuzwhrIxX/s7oPz5XWgCWRD2gXcoqp9gVHABBHp15wMVy6IpaRw3+C9i3+Kw+O2TuSaxdEkp/gnMnv+rjAyVlmR1yvK7GzNiCQppZrFP8Xv1VsSTXJKdX3Z1ImUu4lYXUbZ0QnWBqcNjbYTubCY0jHWttIxCUQuKAYgYlkJNV0iqOlqBUr2xDp8NnArlyVT8qeaxKjDdjBnplULmDOzM6MPswx1VaWD1SuSqK72z18jqX0Vw4/MZ9bHKXu2Lfk1ce85XBZHcoo/wi0KleXWf8PhUBwORVVQhahoa51nVKybvBzfY+/Wx55yfbi3XKPG5jHns/YAzPmsPaPH5flFa/9ls3658q5NvPZoFwiyMWktBi6QQWeygWzv5xIR+QNIw4ppGBCOPSuPH79M8Hu+7dOq6NGvnLVLY/bVOzOXH79KbFKejp3VuOMcJL6wjbAtlVR3j6Tg4g7Yi1x4Eixj5ElwYi+2QhE6sy1D2vZfm7AXuyg7tA0lpzY93mebhEoK8qwQgQV5EcQnBCam65WTNvDfx7sRGb3/xeTH/m0HP870T9xSm0159pOldOhcwVfvpLJ2eSyvTO7Ov15bxWW3b0JscMvZA/2ideWdfy1Xm6RqCnZZNamCXWHEJ/rnYQvesn22jA6dK/nq7RTWLotl5NH55O4MZ9OaaL/p+IIiuFvJIENQjlJEugJDgPn7+e0KEfldRH6v0abfZGdfm43bJXz3adMMTl1ERLm5+6X1vPJgZ8pL99Yez56w3dL7rGkxKcWjhG2qoPS4JHY81gsNtxG3uzm6P9xK+Joy8q7vxM4HexA1v4jwFUH0GtIERhyZR2G+k4zV++/zOuvKTNxu4fsv2/lFz+MRrj1tCBccOYLeA0vp0quME8/JZsrD3bhwzAimPNyNGx9a32ydEWPyKMwLI2OVf/ryfMHjEa49ZTAXHD6M3gNL6ZpextnXbOOtpzsF7Rj2OR7/RLYPOAEfZBCRGOBj4EZVLf7z76o6BSuaNXG2xCYFaR13eh4jxxYx6Zze4MeTand4uOel9Xz/eRK/zNprOMf9bRcjjy5g0nl9mqznSnTiTnJS3cvqwykfHU/cp7twxzuwFdTgSXBiK6jBHWddIneSk8p+0Xi83yuGxhK2sYKqg2Lq1KiPwoIIEpKsWlxCUiVFBf5putWm39BiRh2Vx/Aj8nGGe4iKdjPx32t4/PY+jD11ByOOzOPOSwbiz2sGUFbiYPn8eIYdXsC4/8vh5Ye6A/DTjGRu/FdGs/PvN7SYUUfnMfzIfJxhHqJi3Ex8dA2FeWEktLVqcQltqynK93+H/+6yjR6XT0rHSl78chlg9cU999kybvz7QApyA9sfp2aQwUJEnFjG7W1V/SQQGgcfWcQZV+/g/kt7UFXpz+IoN/17E5kZkXzyWupevSMKOeOqbO6/vDdVlfZ69q8fT4ITV5ITR5ZVa41YUUpNx3AqhsURM7cAgJi5BVQMtyKxVwyKJSyzEqnygFuJWF1GTcemG6V5v6QwbrzVuT9ufCbzfk5pYI/G8/pT3bjw6FFcfMxI/n1LX5bPb8Pjt/fh4MPyOeOybTwwoX+zzmFt4hNqiI61mvNh4W6GHFLI1o1R5OWEcdCIIgAGjyoia3NEs7Vef7IbF44ZycVjR/DvW/pY5bqtD/O+S2TcaTsBGHfaTuZ965+I8/GJfy3bhtXRnDNqBP846mD+cdTB5O4I57rTBgXcuO3G6t9s+HWgCVgNTiyPga8Bf6jqk/7Ic9JzGxk4uoS4BBdvzV/OtCc7cNaEHTjDPEx+22p6rFkSzXN3dmm2Vv9hpYz7Wx6b1kTywtcrAXj9sY5cfd8WnGHK5LfW7tW7u1uTNAou6UDSs1sRl+JqH0beNR0RVZKfzCT6uwLcyU5yb7YGAjTGTvFJybSflAEClUNiqTw4zied2+5dyMAhucTFV/PmRzOZNrUPH77dmzseWMCxJ25h185IJt87Yk/6qe/PIirahcPhYfRh2dx1yyFs3eKbli9cfXcGTqeHh15bAcDaZXE8/0CvBvaqn4R21Ux8ZB02uyICP81MZsHcRMpK7Fx550bsDqW6ysaz9zZPpz4+/E8n7njqD479+w52ZYcz+ca+fsk3oW01Ex/NwGZTxKb8NCOZBd/7tyumcbSMAQRfENUmtQobzljkMOAnYAWw2xXrnao6va594myJOspxXECO5y/Y/VNz8IUt04xHX39gPPo2n3nlX1Hkzm1W4WJ6p+qA5/7hU9r54x9ZpKrDmqPXHAI5ivoz/u5cMRgMBxxVcHtax63dqlYyGAyGlkFLGCH1BWPgDAZDo1BoEQMIvmAMnMFgaCStZ5DBGDiDwdBoAjQ26XeMgTMYDI3GNFENBkNIYo2ito61qMbAGQyGRtNamqitwwwbDIYWhb+WaolIGxH5SETWeH1HjhaRRBH5RkTWe98TaqW/Q0QyRGStiDS4KqBl1eA0eDO6RYJn27vdVR40LYD1l/t/XWld9Ho8iKsmOvtnbadPVDTNz1+TyfKPo9ZgoPh1nekzwExVPV1EwoAo4E7gW1V9REQmAZOA273+JM8G+gMdgDki0ltV6zQapgZnMBgajfr4qg8RiQOOwFqzjqpWq2ohcCrwhjfZG8Bp3s+nAu+papWqbgIygBHUgzFwBoOhcSioR3x6Acm7/T16X1fUyqk7sAuYKiJLRORVEYkG2nsd5u52nLvbYWAasLXW/tu82+qkZTVRDQZDq6ARTdTcehbbO4ChwHWqOl9EnsFqjtbF/kTrrSiaGpzBYGg0qr69GmAbsE1Vd3v6/gjL4O0UkVQA73tOrfS1XRh3BLbXJ1BnDU5EnqMe66iq1zd09AaDIfTw11pUVd0hIltFJF1V1wJjsWK2rAYuAh7xvn/u3eUL4B0ReRJrkKEXsKA+jfqaqL838/gNBkMoovgzitd1wNveEdSNwMVYLcsPRORSIBM4A0BVV4nIB1gG0AVMqG8EFeoxcKr6Ru3vIhKtqmXNKYnBYAgN/DXRV1WXAvvroxtbR/qHgId8zb/BPjjvxLvVwB/e74NE5EVfBQwGQ6jh2wiqtgCnmL6Moj4NHIfV/kVVl4nIEYE8KF+JjnNx0+Nb6ZpeiSo8eUtn/ljknxiRNz22iZFHF1KY5+SqY60o73c8n0HH7pUAxMS5KS22M+GEpkWAv/G2RYwYvYPCwnCuuXiclWdsNXfct4B2KWXk7Ijm4ftHUFoaht3u4YZbF9OzdyE2u/LdrM588E56ozVt4uHTEz5mZ3k0V3x/AgAXpK/g/PSVuNXG3KzOPLp4NA5x89DoH+ifmIvD5uHTjb15ZeXQJpVz6vSfqSi343YLHrdww7kj6Z5ewrV3/4EzzIPHLbwwuQ/rVsY3Kf+bblrAiJHbKSwM5+qrjrfKdOEKRo/OwuMRigrDeeKJkeTnRxIbW8Vdd/9K7975fPNNV1568eBGad14u/eaFfzpmt2/gHYp5eTsiOLh+6xr1i6ljFfenMO2TCu04NrVCTz/5JAmlTE5tYqJj2WQkFyDKsx4rz2fv5FKtz5lXPfgRiKi3ORkRfDozT0pLw3SxIhWslTLp7Ohqlv/5KO+weUGIhIB/AiEe3U+UtX7mnKQdXH1P7P4/fs4/nVFNxxOD+GR/vPZ/82HyXz5RjsmPrlpz7aHr+255/Pld2dSVtz0uA5zZnbhy0+7c8udi/ZsO/PctSxd3JYP3zmMM85dyxnnrmPqlAEcPiYLZ5iHay4ZR3i4i5ffmMPc7zqSs6NxxvyiPivYUJRAjNOapT+yfRZjO23m5K/OpNpjJzGiAoDju2wkzO7mpK/OJMJew4xT3uerTT3JKmta4JlJlx1MceHeaE+X3LSed17uzu+/JDPssFwuuXE9ky5rmtv+b77pyhdf9mTixL0hdz/+qA9vvXkQAKecuo5zz1vF888No7razltvDqBLlyK6dC1qtNacGV348pMe3HLn3u7pM89bx9JFbfnwnXTrmp23jqmvWA+97Kxorrvs6CaVqzZul/Cfh7uwYVUMkdFunv1sOUt+iefGyRt49ZEurFgQz7Gn5/D3y7bz1tOdm63XINp6vIn4Mk1kq4gcAqiIhInIRLzN1QaoAo5W1UHAYGC8iIxq+qHuS1SMm4NGljHzXSu6kKvGRlmx/55eKxfEUlJYV37KESfmM/eLpi8dWrk8mZKSfUO8jTo0mzkzrT/onJmdGX2YNQKuChERLmx2D2Hhblw1NsrLGhdzMyWqlDFpmXyQsTfS07m9VzFl5RCqPZahzq+M9JYOohwu7OIhwu6mxmOntMZ/4ehUISrGCoMXHeMif1fTwx+uXNmOkpJ99y8v33tuIiJce2obVVUOVq1qS3VN0x5M1jXb97xb18yK4jZnZhdGH5bdpLzro2BXGBtWWfFvK8rsbN0QSVL7ajp2r2TFAuuhs/iXeA4bH7wAQH5ZyhAEfLEIV2GtF0sDsoBZwISGdlIrXNfuhYpO78tvRU7pUkVRnoNbnsqke79K1i+P5KV706iqCHy0rAEjSinIdbLdDzE2a9MmsYqCfMvIFORHEp9gxUz9+Yc0Rh2WzdsfTyc83M2UFwZSWtI4g3PXsF95dPEoop1711h2iytiWLtsbh6ygCq3nUcWjWZFXjtmbunO2E6b+fX0N4lwuJj8+yEUVTetrAr86+UlVtPqozRmftyRKY+m8+BLi7n05vWIDSZe6P+gSxddtJyx4zZTVuZk0u1H+T3/3bRJqKIg3zo3BfkRe64ZQEpqOc+9+h3lZQ7efK0fq5YnN1uvXVolPfqVsXZZDJvXRTJqXAHz5iRy+PF5JKdUNZyB3wiRGpyq5qrqearaXlXbqur5qprnS+YiYheRpVgT9b6pNaGvdpordi/jqMH3C2S3Q8+DyvnqzWQmHJdOZbmNs67NaXhHPzDmlLxm1d4aS3rfAjxu4fy/n8DF5xzH385cT0qq7wPaR6VtIa8yglX5bffZbrd5iA+v4vQZ/8e/F43imSO+AZSByTl4VDj0ows46tPzuKTvMjrFFDfp2CdeNJzrzx7JvROGcNJZ2xgwtIATztzGfx7rzUXHHc5/HuvNDff70iBoHG+8MZALLziF77/vwsknNz+afWPJz4vgojOP47rLjuY/LxzEbff8TmRUTbPyjIhyc/cL63jlX10pL3Xw1KSenHz+Dp79bDmR0VbNPmh4fHwdYHwZRe0uIl+KyC4RyRGRz0Wkuy+Zq6pbVQdjzTgeISJ/6ZFX1SmqOkxVhznxvamSm+1kV7aTtUusfqifv25Dz4MqfN6/qdjsyqHjC/jxS/8H3i3MDych0SpDQmIFRQXW+RgzdiuLFrTH7bZRVBjB6pVJ9Eov8Dnfoe12MLbjFr7/v2k8ffgcRqVs5/FDv2VHWQyzMrsBwvK89qgKieGVnNwtgx+zOuNSO/mVkSzelcKApKY9PHY3P4vyw/jtu7b0HlDMuJO388u31vLCn2a3I31A4/vDfGXu91049LCtDSdsIoUF4SQkWgNPCYmVe66Zq8ZOSbH1OWNdAtlZ0XTs1HTPK3aHh7tfWMv3XyTz62zr4bptYyR3/aMf1582kB++TCY7s+lN/Uaxex6cL68DjC8m/x3gAyAVa/bwh8C7jRHxegiYC4xv3OHVTcEuJ7nbw+jYw/pzDT6shMx1gb/AQw4rZuuGSHJ3+K9Pajfzfk1l3PhMAMaNz2TeL6kA5OREMmhoDqCER7jo0y+frd7ROV94YslIDv/kAo769Hxu/Gkc83Z0YOIvY5mztSujU6x+vq6xhThtbvKrIsgui2F0ShagRDpqGJycw8aihPpF9kN4pJvIKNeez0NG57MlI5q8XeEcNMwy0INGFJCVGdXovOujQ4eSPZ9Hjcpi29amDY74wrxfUhg3fgsA48Zv2XPN4uKrsNmsHpmU1DI6dCwle3tTR/iVGx/ewNaMSD79b4c9W+MTrRqhiHL2hG1Mfzd4brL8tFQr4PjSByeq+lat79NE5NoGdxJpC9SoaqGIRALjgH838Tj3ywv3pHH7c1twOJUdmWE8cbP/RpAmPbuBgaNLiEtw8da8pUx7Ko1Z77dlzMl5zP2i+bW32+5ZwMDBu4iLr+bND6czbWo/PnynN3fct4BjT9jMrp1RTL5/JABffdaDm25fxEtT5yAC38zowuaNTZtWUZuPNvTh4dFz+frk96lx27nt16MBYdraATxyyPdMP/kDBPh4QzprCxvfJE9IrOLup5YDYHcoc6ensOjXZCr+6eDK29Zitys11Tae+2ffBnKqm9sn/cbAgTnExVXx1ltf8Na0AQwfnk3HjsWoCjk7o3nuub3TQV5/40uiolw4HB4OGZ3FXXcdSWamb+fytnsX1rpmM5g2ta91ze5fyLEnbrGu2X2W956DBuVy/iV/WNNjPMLzTw5udL/pbvofXMK4/8tl05oonv9iGQBvPNGZDl0rOel8y4/cr7MTmf1R2/qy8S8twHj5gmgdZlZEdt/FtwGFwHtYxToLCFfVB+vNWGQgli8nO96lF6r6z/r2iZNEHWkb15jjbzLiaNwoZHOwdevUcCI/sv7Sdg0n8hO9Hg9e/5amBq/fU4Ls8FKD5PByXsXXFLlzm9V2DO/aUVPuvsGntJmX37aoHm8iAae+GtwiLIO2+2RcWes3Beo1cKq6HGjazEaDwdCikVZSg6tvLWq3YB6IwWBoJahAC1iG5Qs+zYz1jn72A/ZMhlLVNwN1UAaDoYXT2mtwuxGR+4AxWAZuOnA88DNgDJzB8L9KKzFwvkwTOR3LdckOVb0YGASNmLBmMBhCjxBaqlWhqh4RcXmj4ORgBYswGAz/i/jX4WVA8cXA/S4ibYD/YI2sltKAm2CDwRDatPpR1N2o6jXejy+LyEwgzjsFxGAw/K/S2g2ciNTp4VBEhqrq4sAcksFgaOmEQg3uiXp+U6D5nvz2m3Nwzpy6mufZoTF4NgVusff+6PmA/32S1cXfF20MmtZHBwev61c9wXWF4akKjqsjv5WrtffBqWrgnGgZDIbWSwsZIfUFE9neYDA0HmPgDAZDqCItwJmlLxgDZzAYGk8rqcH54tFXROR8EbnX+72ziIwI/KEZDIaWiKjvrwONL0u1XgRGA+d4v5cALwTsiAwGQ8unlbgs96WJOlJVh4rIEgBVLRAR//vrNhgMrYcWUDvzBV8MXI2I2PEWyeuKvJV0MRoMhkDQEpqfvuCLgXsW+BRoJyIPYXkXuTugR2UwGFouGkKjqKr6togswnKZJMBpqur/QJaN5OYnMxk5roTCXAdXHp0eFE2bTXluxjrydji59yL/zaq/6bFNjDy6kMI8J1cduzey4in/2MkpF+bgdsOC79rw2sPNj+2QnFrFxMcySEiusYIxv9eez99IZdIz6+jYzQpZGBPnprTYzrWnDGqSxtrXo9j0USQIxPd2MWJyESufjWX79+HYnEpMJzfDJxcRFqd4amDhPfEUrnbgcQtdT62g7xW+x3xtqFzd+pRx3YMbiYhyk5MVwaM396S8tPmTB5JTq5j4+AYS2tagHmHGe+34/PW9Ua3+flk2l92ZyVkHD6W4wP/xPwL1X/SZUKnBiUhnoBz4svY2Vc30RcDbvP0dyFLVk5p6oH9m9vuJfDE1mVufCd4yqNMu28XW9eFExfr38fXNh8l8+UY7Jj65ac+2gaOLGX1MIVeP709NtY34JP8sLXO7hP883IUNq2KIjHbz7GfLWfJLPI/c0HtPmsvu2Ex5ib1J+ZfvtJExLYrjvsrFEQG/3hRP5vRI2h9SxUE3lWBzwLLHY/hjSjSDJpaydVYEnmo47os8XBUw86S2dD6xkug0t1/KdePkDbz6SBdWLIjn2NNz+Ptl23nr6eZHX3O7hP9M7sKGVdGW3hcrWfJzHJkZUSSnVjHksCJ2ZgWuqzpQ/0WfaSUGzpdR1K+Br7zv3wIbgRmN0LgB8HuNb+X8GEoKgjeNLzm1mhFji5nxrv8jO61cEEtJ4b5lOen8HD54MYWaausSFeX5pxZQsCuMDatiAKgos7N1QyRJ7WtHkFKOOCGPuV8mN1nD4xbclYLHBe4KIbKdm5RDq7F5i5g0qIaKnZYBFQFXhTdtpWBzKo7oxt+0dZWrY/dKViyw4qIu/iWew8bnN7lcf9WL3quXEUFSivUQuvLuLbz2SKeAGYFA/hd9JWSmiajqQao60PveCxiB5bK8QUSkI3Ai8GrzDvPAc9UDWbz6rw5okB6Yad0q6T+ilKc/W82j76+h98CmR0Wvi3ZplfToV8baZTF7tg0YXkJBrpPtWyKblGdUew/pF5fx9di2fHlEO5yxSsqh+4bg2/RJJKmHW4vLOx5biSNS+fKIdnw1ti3pl5QR3qZ5d0btcm1eF8mocVaQ6cOPzyM5xf+L2tulVdGjfzlrl0YzcmwBuTvC2LSmqUGeGybY/8XWjC81uH3wukka7mPyp7HiqtZ5KUTkChH5XUR+ryE4HhUay8hxRRTmOshY4d8I7PVhd0BsvIsbT+vLq5M7cueLG/BnlSAiys3dL6zjlX913adPasxJufzwVdNrb9VFwvbvwjnhm12c/EMOrgphyxd7YhWx+uVobHbofHIlAPkrnIgdTv4hhxO/yWXd1GhKtzateby/cj01qScnn7+DZz9bTmS0G1dNo//yDeu9uI5XHuyC2yWcPSGLt57u6FeN2hyI/+J+CRWX5SJyc62vNmAosMuH/U4CclR1kYiMqSudqk4BpoAV+LmhfA8E/YaVMerYYoYfvYqwcCUq1s1tz27h0eu7BEwzN9vJLzMTAGHdshg8HiE+0UVRfvObqnaHh7tfWMv3XyTz6+y9zRybXTnkuHyuP+2gJue987cwotPcRHgvZcdxleQuCaPLKZVs/iyC7LnhHDk1H/HOAd3yVQQph1Vhc0JEkoekodUUrHQS06lxfXB1lWvbxkju+kc/ANK6VjBiTEGTy7ZfvRfXW3qzEumaXk5Kxype/HoFAMkp1Tz35UpuPK0/Bbn+6Y87EP/FvxBKo6hAbK3PLqy+uI992O9Q4BQROQEr3GCciExT1fMbf5gHlqmPdGDqIx0AGDi6hNOv2hXwP9SvsxMYdEgJy+fFkdatEqfTQ1G+P/oclRsf3sDWjEg+/W+HfX4Zcmgh2zZGkLuj6TGFolI95C1z4qoAewTsnBdO4oAasn8KY82rMYx5Mw9H5L7pc+ZbBtBdIeQvC6P3heV+K1d8Yg1F+U5ElLMnbGP6uyn15NFIvUc2sXVDJJ++lgrA5rVRnDPi4D0pXv9xCdefOsCvo6gH4r+4X1pkVeSv1HvHeEdAY1T11sZmrKp3AHd48xkDTPSncZv04hYGji4lPtHFtN9X89YT7Zl1ADtdm8OkZzcwcHQJcQku3pq3lGlPpTH7g2RufmwTL89eiatGePyW7lizdJpH/4NLGPd/uWxaE8XzXywD4I0nOrPwhwSOPLF5gwtgDSB0PK6Kb/6ejNiVhL4uup9ZzqyTk3FXCz9emghA4qAaht1fTM9zy1l4VzyzTk4ChK7/V06bdJffytWhayUnnb8DgF9nJzL7o7bNKt8evWGljPtbLpvWRPL8V1aN7Y3HO7Fwbhu/5N+SEfw7gPDnmRYikgi8D3QFNgNnqmqBN+0dwKWAG7heVWfVm7fW4UFXRByq6hKRb1V1bDMLMAbLwNU7TSROEnWkNEvKdyR46+TE4f95UPXqOYM3unx6iHr0JUQ9+s73zKFY85v154/s0Em7XnpzwwmBNf+6eZGqDqsvjbcbbBhWvJeTRORRIF9VHxGRSUCCqt4uIv2Ad7EGOjsAc4Deqlpnf0Z9Pa67I2ctFZEvROQCEfnb7pdPpfOiqnP9OQfOYDAcQPzoTaSOmRanAm94P78BnFZr+3uqWqWqm4AMLGNXJ7486hOBPKwYDIpVQ1XgEx/2NRgMoYjvFdxkEfm91vcp3oHF3TyNNdOidl9/e1XNBlDVbBFp592eBsyrlW6bd1ud1Gfg2nmrjivZa9h200q6GA0GQyBoRB9cbl1NVF9nWtTeZT/b6j2S+gycHYhpSqYGgyHE8Y8F2O9MC2CniKR6a2+pQI43/Tag9oLsjsD2+gTqM3DZqvrPph+7wWAISfw0ibeumRYi8hhwEfCI9/1z7y5fAO+IyJNYgwy92DtWsF/qM3AH3h2nwWBokQR4nekjwAcicimQCZwBoKqrROQDYDXWnNwJ9Y2gQv0GLkjzNQwGQ6vDzwZOVecCc72f86jD/qjqQ8BDvuZbX+Bn/7hdMBgMIUcoLdUKLkGagBvMybcSFuSJvmn+Wo7UMJ8cFdtwIj+x5U3/rELwha6TGu90sznI5iD5NfRHzauFLKT3hZZn4AwGQ4tGaD0d9MbAGQyGxmNqcAaDIVRpCd56fcEYOIPB0HiMgTMYDCFJiDm8NBgMhn0xNTiDwRCqmD44g8EQuhgDZzAYQhVTgwsCb8xbRUWpHY/HijR+3Qnpfsv7psc2MfLoQgrznFx17AAAuvUt5/rJm4mI8rBzWxiP3tCD8tKmh7jbjTPMw2PvrMQZ5sHuUH6emcS0Zztz6e2bGXlUAa4aITszgicn9aSspPGX7MbbFjFi9A4KC8O55uJxAMTEVnPHfQtol1JGzo5oHr5/BKWlYdjtHm64dTE9exdisyvfzerMB+80/bxGx9Rw/b2r6NKjFBCefqA/a5a34eSzMjnprEzcbmHhz22Z+kzvJuUvZW4SX8rCmVkJIuRfk4arQxhJT23FnlODu52T3Js7ozF2cKmVdlMF4oayI9tQ8jffV0fcePtiRhyyg8KCcK75h7VUMia2mjvuX0i71HJysqN4+L7hlJZaEbTOPG8dx564BY9HePmZg1i8sH2TynjTY5sZObaIwjwHVx3TH4DL7tzGyHGFuGpsbN8SzpMTu1BWHKTbWWmMw8sDin+DRP4JEdksIitEZOmfvHr6jdvO6Mk1x/bxq3ED+ObDZO6+aN+b7qZ/b+K/j3Tk6uMG8OusBE6/MtsvWjXVwqQL+zPhlMFMOGUQBx9RSJ/BJSz5pQ1XnTiYa04eTNbmCM66aluT8p8zswv33HbIPtvOPHctSxe35fLzj2Pp4racce46AA4fk4UzzMM1l4zjhiuO4vhTNtEupenLlq64dQ2Lfk3mqr8fxrVnjWbrxmgGDstn1JgcJpx1CNeccSifvNn0qFAJ/82mcnAMO57tzY7He1DTMZzYz3KpPCiGHc/3pvKgGOI+taJcRv1WhNQoO5/sxc5HexDzTT72nOoGFPYyZ2Zn7rn1T+fxvHXWeTz3GOs8nr8egE5dijli7Dauuuho7rl1NBNuXobN1rRqzzcfJnH3hb322bb4pziuPKY/Vx/Xj6xN4Zw1YUeT8m4Ku4POhERkez9wlKoObijwREtj5YJYSgr3fSKmda9kxXxr7eXin+I49Hh/xdgUKsutmqDDoTgciios/rkNHre1KGbN0liSU3y/GWuzcnkyJSX7xuUcdWg2c2Z2Bqwbd/Rhlt9AVYiIcGGzewgLtwIll5c1bS1tZLSLAUMLmP2Z5VXa5bJRVurkhNO38uHUbnuCMBcVNC1MoZS7Cf+jjLKxCdYGpw2NthO5sJiyMW0AKBvThsiFxd4dQKo84Fak2oM6BI30/RZYuSyZkuJ9z8Wow3b86TxaD73Rh+3gx2874qqxszM7mu1ZMfTu27T/i/Vf3LelsPinuL3/jcXRJKfUNCnvJhMqgZ9bNCpMfncDKHw9LYkZbzcv5F1DbFkXyahjCpn3TQJHnFhA29SmGZz9YbMpz362jA6dK/nq7RTWLtt3Efuxp+fww9f+K1+bxCoK8q0ApQX5kcQnWFGdfv4hjVGHZfP2x9MJD3cz5YWBlJY0LWhxalo5RQVh3HT/Krr1LiHjjzheeSydtC7l9B9awIUT1lNdbee1p3qzfnV8o/N37KzGHecg8YUsnJsrqe4RSeHFqdgLXXgSLEPkSXBiL7LCEJaPiidyQQkdLl+DVHko/Ecqntjm3QJtEiopyIsAoCAvYs95TGpbwZpViXvS5e6KICm5olladXHsWXn8+GVCQPKuC6kjGl9LI9A1OAVmi8giEblifwlE5AoR+V1Efq+hcaHTbjqtF9eOT+eu87tzyj9yGTCy1B/HXCdP3tqNky/M4bmvVhEZ7cZV478lxx6PcO0pg7ng8GH0HlhKl157m4VnX70Nt0v4/ovAGnCA9L4FeNzC+X8/gYvPOY6/nbmelNSmNVFtdqVnnxKmf9SR688dTWWFnTMu3ozN7iEm1sXNF43kv0/3ZtK/l9Gkx70bwjZWUHpsIjsf74mG24j1Nkf3R1hGOWqD7VP6kP1iOrFf5mLf6b+HVG2CFZXy7GuzcbuE7z5NbDixv/C19tYCbGCgDdyhqjoUOB6YICJH/DmBqk5R1WGqOsxJ45oq+Tutp3RRnpNfZsTTZ3BTIqL7zrYNkdx1QTrXndSfuV8kkb0lwu8aZSUOls+PZ9gRhQCM+78cRhyVz6O39MKfPhwK88NJSLRqFAmJFXuaiWPGbmXRgva43TaKCiNYvTKJXulNa1rl5USQmxPO2pVtAPjl2/b07FNMXk4Ev37XDhDWrYpHPUJcm8Y3sdxJDtxJTqp7RwFQPiqOsE0VuNs4sBVY+dkKanDHW7W0qJ+KqBwSAw7BE++gKj2KsA3Nq1UVFkSQkFQJQEJS5Z7zmJsTSdt2e/NObltJXm5ks7T+zLjT8xg5tohHr+9GsP17mD44QFW3e99zgE9pIIZhYwiPdBMZ7d7z+eAjS9i81v8GpzbxSdZNI6Kcc912vn7bP/7J4hNriI61mlFh4W6GHFLI1o2RHHx4AWdckcUDV/WlqrL5o7W1mfdrKuPGZwIwbnwm835JBSAnJ5JBQ3MAJTzCRZ9++WzNbJrPt4K8cHbtjCCti1UDHDQij8xN0fz2fTsGDbf8qXboXIbD6aG4sPH9fJ4EJ+4kJ44sq+YfsaKUmo4RVAyLI3puIQDRcwupGB4HgDvZScTKMlBFKj2Er6/A1aFp/X+7mfdLyr7n8eeUPduPGLsNh9NN+9QyOnQsZd0f/mtGHnxkEWdcvYP7L+1BVWUwutL3RTy+vQ40dUa2b3bGItGATVVLvJ+/Af6pqjPr2idOEnWkbZxP+ad0ruK+1zYBYLfD95+14d1nfXf02JDDy0nPbmDg6BLiElwU5DqY9lQaEVFuTr7QCvDzy8wEpv67I748ORtyeNk1vYyJj2ZgsyliU36akcw7z3fitTmLcYZ5KPYOdqxZGsvz9/ZoWO9PDi9vu2cBAwfvIi6+msKCcKZN7cdvP6dyx30LaNu+nF07o5h8/0hKS8KIiHRx0+2L6NylGBH4ZkYXPn6/7ikcUlx/t0D33sVcf+9qHE4PO7ZF8vT9A6issHPj/avo1rsYV42N157uzfKFSQ2Wa9Pzf32gODdVkPhSFrgUV/sw8id0RFRJemIr9twa3MlO8m7phCfWgVS4rf66bZZBLDuqDSWn7v8htT+Hl7fdu5CBQ3Kt85gfzrSpffjtpw7c8cAC2ravYNfOSCbfO2JPn+VZF6zl2BO24HbbmPLcQfw+v+5pIp56HF5Oem5jrf+ik2lPduCsCTus/0aB97+xJJrn7mx4NHqeaxbFnuZFto9O7qT9T7zJp7QL37ylwcj2gSSQBq47Vq0NrMGMd7z+1OukMQauuRiPvn7SasDA+ZP9GbhAEWyPvvUZOH/iFwOX1EkH+GjgFrx1YA1cwEZRVXUjMChQ+RsMhgNIC+hf84XWPU3EYDAEnd0TfVsDxsAZDIZGI57WYeGMgTMYDI2jhcxx8wVj4AwGQ6NpCVNAfMEYOIPB0HhMDc5gMIQqZpDBYDCEJorldqYV0PIMXJBOnLrdQdEBsIVFB00LQGpcQdNSd/A6Y7reGRhvHPtj62P+XTfaEGlnBFWu2Zg+OIPBEJKYeXAGgyF0UTVNVIPBELqYGpzBYAhdjIEzGAyhiqnBGQyG0EQBd+uwcMF3BWowGFo9/nBZLiKdROR7EflDRFaJyA3e7Yki8o2IrPe+J9Ta5w4RyRCRtSJyXEPHaQycwWBoPLtHUht61Y8LuEVV+wKjsOK29AMmAd+qai/gW+93vL+dDfQHxgMviki9vvyNgTMYDI3GHzU4Vc1W1cXezyXAH0AacCrwhjfZG8Bp3s+nAu+papWqbgIyaCDOS6vtg3OGe3jikwycYYrdofz0dRveejxwrrpPuzSH48/JRQRmvJPMp6+187uGzaY888Ei8naGcf+EgUx6fBVp3axIYTGxLkpLHFz39+HN1jnljI0cd8oWRGDWF535/IMeXDJhFSMO3YmrRsjOiubpyUMoK22+q/Wp03+motyO2y143MIN546ke3oJ1979B84wDx638MLkPqxb2fi4qAA33raIEaN3UFgYzjUXW+7uY2KrueO+BbRLKSNnRzQP3z+C0tIwxozL5O9nr9+zb7fuRVx/xdFszGjjk5aUuWnzwnYcmVZMh8JrOxA5r4Tw30vAIbhSwii8rgMabVUqYj7OJWpOAdiEostSqBoS06Qy3vTYZkaOLaIwz8FVx/QH4MJbshh9bBEeDxTmOXjilq7k72xa/NpG0zh3Scki8nut71NUdcqfE4lIV2AIMB9or6rZYBlBEdl9s6UB82rtts27rU4CauBEpA3wKjAA65Rcoqq/+SPvmirhtjN6UFlux+5Qnvwsg4XfxbJmsf+XRXVJr+D4c3K5/qQ+1NQIk6dlMP+7OLZv8m8Ur1Mv2MbWjVFERVtLrR6Z2H/Pb5fdmkFZafMvV5duxRx3yhZuvuxwalw2HnxiHgt/bc+ShW15/eW+eNw2Lr56NWdesJ6pL/Vrth7ApMsOprhw7813yU3reefl7vz+SzLDDsvlkhvXM+myprntnzOzC19+2p1b7ly0Z9uZ565l6eK2fPjOYZxx7lrOOHcdU6cMYO6czsydY0Wh79qtiHse+s1n4wYQ/+oOqobEUHBbJ6hRpNpDVaWH4gvagV2IfXMnMR/nUnJhexxbq4j8uYicZ3tgz3eRdN8Wcl7oCfbGh0P45sMkvnyjHROf2rRn20evpPDmE9a9ferFOZx3Q7ZPQWf8gQDi+yBDbkMxGUQkBvgYuFFVi6XuoLL7+6HeAwl0E/UZYKaq9sGKz/CH/7IWKsutJ6XDqdidGrDJ1Z17VvLHkmiqKm143MLyeTEcOr7QrxpJ7SsZfkQesz5O3c+vyuHH7eKHr5tfa+zUtZS1qxKoqnLgcdtYsTSJ0Udks2RBOzxu6++wZlUCSe0Ct+5TFaJiLCMeHeMif1fTQ/etXJ5MScm+NZdRh2YzZ6ZlyObM7Mzow7b/Zb8jx27lh287+awj5W7CVpdTPq6NtcEpaLSdqsExe4xWTe9I7HlWaMmIBSVUHBYPThvu9mG4UsNwrm/aOV25IJaSwn27mspL936PiHIHfWGBqPr0ajAfESeWcXtbVT/xbt4pIqne31OBHO/2bUDti9YR+OvFrUXADJyIxAFHAK8BqGq1qhb6U8NmU178Zi3vL1/Fkh9jWLskMIvaN6+N4KCRpcS2cREe4WH40cW07dD4QMX1ceWkDP77RA88+1nEPODgIgrznGzPjGq2zpaNsQwYlEdsXDXh4S6Gjc6hbfvKfdIcc2Imi37zTxNcgX+9vIRn3p3P+L9vA2DKo+lcctN63pj1E5fesp7Xn+3pF63dtEmsoiDfWixfkB9JfELVX9IccVQWP3zX0ec87Ttr8MTZafPcdtrevJH4F7YjlfterKhvC/c0Q+15NbiT9ta43UlO7Pn+dYJw0a1ZvDVvOUedls9bT3Twa9714qfI9mJV1V4D/lDVJ2v99AVwkffzRcDntbafLSLhItIN6AUsqE8jkE3U7sAuYKqIDAIWATeo6j7x2ETkCuAKgAgadwN7PMI1x6QTHefmvtc20SW9gi1r/e8FYmtGJB+82J6H311PZZmdTasjcbv8F0l8xJG5FOaHkbE6loOG/zWK/JEn5DB3et0xNRvD1i2xfPR2T/719G9UVtjZlBGH2723LGdduA63W/h+tu83f31MvGg4+bvCiU+s5qGXF7NtUzSHHpPDfx7rzS/ftufwY3dyw/1/cNeVQ/2i5wvpffOpqrKzZZPv/X7iVpwbKym6PIWa3lHEvbqDmE9yKTnXehDEfLgLtQsVR3rz3N/N7efg8288lsYbj6Vx1oRsTv7HLqY9GSwj57e1qIcCFwArRGSpd9udwCPAByJyKZAJnAGgqqtE5ANgNdYI7ARVrdctUCCbqA5gKPCSqg4ByvAO99ZGVaeo6jBVHeakaU2VsmI7y36LYfhRJc064PqY9V4y1x7fl4mn96ak0E7WpuZFRK9NvyHFjBqTy9TZv3H746sZOLKQiY+sBsBm93DIuF38ONN/MUFnf9WFGy45ktsnHEZJcRjbt1o137HHZzL80J08/sBQ/HU37m5+FuWH8dt3bek9oJhxJ2/nl28tw/DT7HakDyjyi9ZuCvPDSUi0moMJiRUUFex7rY44ehtzv22cAXcnOXEnOanpbT2EKw+JxbnRqvlGfldIxO+lFN6UBt7+I3eyE3ve3hqbPa8Gd0Jg6hPff5bIYcf/9cEYSPw0ivqzqoqqDlTVwd7XdFXNU9WxqtrL+55fa5+HVLWHqqar6oyGjjOQBm4bsE1V53u/f4Rl8PxCfKKL6DjLeIdFeBh6eClbM/zb6b+PXpLVJG3boZpDjy9k7ucJDezhO68/3Z0Lxx7CxceO5t8T+7F8fhsen2R18A8ZXcC2TVHk7fRf2eLbWE22tu3LOeTIbH6Yk8bBI3M4/bwM/nn7CKqq/HMjhke6iYxy7fk8ZHQ+WzKiydsVzkHDrBty0IgCsvzQ9K7NvF9TGTc+E4Bx4zOZ98vefk0R5fAx2/jxO9/73wA8CQ7cyQ7sWda5C19ehqtjOOGLS4n5NJf8Ozuh4Xtvp8rhMUT+XAQ1Huw7q3FkV1PTy3+tiw5d93YrjDqmiK0bAvff3y/+mQcXcAIZ+HmHiGwVkXRVXQuMxapa+oXE9jVMfCYTmw1sNvjxy3jmz4nzV/Z/4d4pG4lNcON2Cc/f1YnSouDMsDni+Bx+mO7fKSl3Tl5IXFw1LpeNl544iNKSMK66eTlOp4eHnrYGudesSuCFx5oXtzshsYq7n1oOgN2hzJ2ewqJfk6n4p4Mrb1uL3a7UVNt47p99m6xx2z0LGDh4F3Hx1bz54XSmTe3Hh+/05o77FnDsCZvZtTOKyfeP3JN+wKBccndFsiO78f21RZenkvBUFuJSXO2tKSFtb92I1ChJ928BoLp3FEVXp+LqHEHFIXG0u24D2IWiy1OaNIIKMOm5jQwcXUJcgou35i9n2pMdGH5UER17VKIeYWdWGM/d0blJeTcJbdQo6gFFNIBWVkQGY00TCQM2Aherap116ThJ1JEyNmDHsw+2eidA+xV7fOAM7/6QNsHT09LyoGmRELxybf23/7ogfCHtjHVB0ZnnmkWxJ79Z/Q9xMWk6cuDVPqWd89s9ixqaJhJIAloNUdWlwAErnMFgCAy+TAFpCbTalQwGg+EAYgycwWAISRQwQWcMBkMoIvi2SqElYAycwWBoPPtbctMCMQbOYDA0DtNENRgMoYxpohoMhtDFGDiDwRCatIxlWL7wv2vgPPU6IfAr7oLgLoQm2HrBYteuoEl1+L+gSQEwc/vSoOiMOM4PK09aUVSt/10DZzAYmozpgzMYDKGLMXAGgyEkUcBjDJzBYAhJzCCDwWAIZYyBMxgMIYkC7taxlMEYOIPB0EgU1Bg4g8EQqpgmauAZNqaYqx7cjt2mzHg3kQ+e909oPaPV+rWCredvrU9fTWbG20mowvHn5fO3y3fx0JVd2OYNLlNWbCc6zs1Lc9YCsHF1BM/e3omyEhs2Gzw3fR1hEQEyQmYUFUQkHXi/1qbuwL2q+rQ/8rfZlAmTs7jj7O7kZjt5bvp65s2KJ3O9/6MLGa3WpRVsPX9rbV4TwYy3k3j263U4w5Q7z+3ByLFF3PXKlj1pXnmgA9Gx1moctwseva4Ltz67hR79KynOt2N3BtgAtZIaXMDCBqrq2t2xDoGDgXLgU3/lnz6knO2bw9iRGY6rxsbcz9sw+jj/xtc0Wq1TK9h6/tbKXB9O36HlREQpdgcMHF3KLzPa7PldFX78og1HnWYtyVv0Qyzd+lbQo78VSjAu0Y090DGVWknYwEDGRa3NWGCDqm5pMKWPJKXUsGt72J7vudlOklNr/JW90WrFWsHW87dW1z6VrJgfTXG+ncpyYeF3ceza7tzz+8r50SS0dZHWvRqAbRsjEIE7z+nOhGN788EL/g0z+RdUwe327XWACVYf3NnAu/v7QUSuAK4AiMD3AMCyn8BngXpgGK3WpRVsPX9rde5VxZnX5HDH2T2IiPbQrV8FdsfeDL//LIExp+11qOB2wcoF0Tw3fR3hkR4mndWTXgPLGXJ4adMPoiFaQO3MFwJegxORMOAU4MP9/a6qU1R1mKoOc+J7LMrcbCdtO1Tv+Z6cWkPeDmc9ezQdo9W6tIKtFwit8efm88LsdTzxaQaxbdykdasCLGP2y/R4jjylcE/atqk1DBxdRnySm4goZfjRxWSsiGyWfoOYJuoejgcWq+pOf2a6dmkUad2qad+pCofTw5hTC5k3O96fEkarlWoFWy8QWoW5VuMqZ5uTX6bHM+a0QgAW/xRLp55VtO2wtwl88JgSNq2OoLJccLtg+W8xdO5d1Sz9+lFrFNWX1wEmGE3Uc6ijedocPG7hhbvSmPzORmx2mP1eIlvWBWZEzmi1Lq1g6wVC65+XdaWkwIHdqVw7eRuxbaz+rB8+37d5ChDbxs3frtzFdSf0RgRGHF3MyHHFzdKvFwVtJRN9RQNYjRSRKGAr0F1VGxxWipNEHSljA3Y8BkNLZVbQHF5u5fdllfvpNfSdeEdbHR17qk9pZxW+tkhVhzVHrzkEtAanquVAUiA1DAZDkFE1YQMNBkMI0wIGEHzBGDiDwdBo1NTgDAZDaNIypoD4gjFwBoOhcZjF9gaDIVRRQFvAMixfCNZaVIPBECqo1+GlL68GEJHxIrJWRDJEZJK/D9XU4AwGQ6NRPzRRRcQOvAAcA2wDForIF6q6utmZezE1OIPB0Hj8U4MbAWSo6kZVrQbeA3ybQewjAV3J0FhEZBfQWJdKyUBuAA7nQGsFW89o/W9odVHVts0RFpGZXn1fiAAqa32foqpTvPmcDoxX1cu83y8ARqrqtc05vtq0qCZqU068iPwerKUgwdQKtp7RMlq+oqrj/ZTV/paM+bXGZZqoBoPhQLEN6FTre0dguz8FjIEzGAwHioVALxHp5vUbeTbwhT8FWlQTtYlMCVGtYOsZLaMVVFTVJSLXArMAO/BfVV3lT40WNchgMBgM/sQ0UQ0GQ8hiDJzBYAhZWrWBC/Qyj1o6/xWRHBFZGSiNWlqdROR7EflDRFaJyA0B1IoQkQUissyr9UCgtGpp2kVkiYh8FQStzSKyQkSWisjvAdZqIyIficga77UbHSCddG95dr+KReTGQGiFAq22D867zGMdtZZ5AOf4c5lHLa0jgFLgTVUd4O/8/6SVCqSq6mIRiQUWAacFqFwCRKtqqYg4gZ+BG1R1nr+1amneDAwD4lT1pEDpeLU2A8NUNeCTb0XkDeAnVX3VOyIYpaqFAda0A1lYk2P9FnM4lGjNNbiAL/PYjar+COQHIu/9aGWr6mLv5xLgDyAtQFqqqruDZzq9r4A98USkI3Ai8GqgNA4EIhIHHAG8BqCq1YE2bl78HlA91GjNBi4NK6DNbrYRIENwoBCRrsAQYH4ANewishTIAb5R1YBpAU8DtwHBcgerwGwRWeQNMB4ougO7gKne5verIhIdQL3d1BlQ3WDRmg1cwJd5HEhEJAb4GLhRVQMWA05V3ao6GGsW+QgRCUgTXEROAnJUdVEg8q+DQ1V1KFZs3gneroZA4ACGAi+p6hCgDAhYnzA0HFDdYNGaDVzAl3kcKLz9YR8Db6vqJ8HQ9Dap5gL+Wmf4Zw4FTvH2i70HHC0i0wKkBYCqbve+5wCfYnVrBIJtwLZatd+PsAxeIAlIQPVQozUbuIAv8zgQeDv+XwP+UNUnA6zVVkTaeD9HAuOANYHQUtU7VLWjqnbFulbfqer5gdACEJFo7yAN3ubisUBARsFVdQewVUTSvZvGAn4fFPoTAQmoHmq02qVawVjmsRsReRcYAySLyDbgPlV9LRBaWDWdC4AV3r4xgDtVdXoAtFKBN7yjcTbgA1UN+PSNINEe+NR6XuAA3lHVmQHUuw542/uw3QhcHCghb0D1Y4ArA6URKrTaaSIGg8HQEK25iWowGAz1YgycwWAIWYyBMxgMIYsxcAaDIWQxBs5gMIQsxsC1IkTE7fUgsVJEPvROF2hqXq97oxrhXVrUr560Y0TkkCZobBaRv0Rfqmv7n9KU1vf7ftLfLyITG3uMhtDGGLjWRYWqDvZ6NKkGrqr9o3c+W6NR1csa8FYyBmi0gTMYDjTGwLVefgJ6emtX34vIO1iTg+0i8piILBSR5SJyJVgrJETkeRFZLSJfA+12ZyQic0VkmPfzeBFZ7PUR9613wf9VwE3e2uPh3hUQH3s1ForIod59k0RktnfB+Svsf73wPojIZ97F8Kv+vCBeRJ7wHsu3ItLWu62HiMz07vOTiPTxy9k0hCStdiXD/zIi4sBai7h7Zv4IYICqbvIaiSJVHS4i4cAvIjIbyytJOnAQ1iz/1cB//5RvW+A/wBHevBJVNV9EXgZKVfVxb7p3gKdU9WcR6Yy1mqQvcB/ws6r+U0ROBHzx4HGJVyMSWCgiH6tqHhCNtdbyFhG515v3tVjBVq5S1fUiMhJ4ETi6CafR8D+AMXCti8hay7d+wlqzegiwQFU3ebcfCwzc3b8GxAO9sPyVvauqbmC7iHy3n/xHAT/uzktV6/KBNw7o510GBRDnXfd5BPA3775fi0iBD2W6XkT+z/u5k/dY87BcKr3v3T4N+MTrYeUQ4MNa2uE+aBj+RzEGrnVR4XVttAfvjV5WexNwnarO+lO6E2jYnZT4kAasro3Rqlqxn2Pxee2fiIzBMpajVbVcROYCEXUkV69u4Z/PgcFQF6YPLvSYBVztdbmEiPT2etP4ETjb20eXChy1n31/A44UkW7efRO920uA2FrpZmM1F/GmG+z9+CNwnnfb8UBCA8caDxR4jVsfrBrkbmzA7lrouVhN32Jgk4ic4dUQERnUgIbhfxhj4EKPV7H61xaLFSTnFaya+qfAemAF8BLww593VNVdWP1mn4jIMvY2Eb8E/m/3IANwPTDMO4ixmr2juQ8AR4jIYqymcmYDxzoTcIjIcuBBoHYsiDKgv4gswupj+6d3+3nApd7jW0WA3NQbQgPjTcRgMIQspgZnMBhCFmPgDAZDyGIMnMFgCFmMgTMYDCGLMXAGgyFkMQbOYDCELMbAGQyGkOX/AU7RBT68J7ESAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(ground_truth, predicted_classes, labels=classes_idx)\n",
    "disp = ConfusionMatrixDisplay(confusion_matrix=cm)\n",
    "disp.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Calculer les métriques de performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 79.64%\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>F1-Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Another party (please specify)</th>\n",
       "      <td>97.24%</td>\n",
       "      <td>100.00%</td>\n",
       "      <td>98.60%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Bloc Qu&lt;e9&gt;b&lt;e9&gt;cois</th>\n",
       "      <td>92.78%</td>\n",
       "      <td>99.13%</td>\n",
       "      <td>95.85%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Conservative Party</th>\n",
       "      <td>74.81%</td>\n",
       "      <td>67.71%</td>\n",
       "      <td>71.09%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Don't know/ Prefer not to answer</th>\n",
       "      <td>58.79%</td>\n",
       "      <td>50.76%</td>\n",
       "      <td>54.48%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Green Party</th>\n",
       "      <td>78.48%</td>\n",
       "      <td>88.75%</td>\n",
       "      <td>83.30%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Liberal Party</th>\n",
       "      <td>66.45%</td>\n",
       "      <td>64.00%</td>\n",
       "      <td>65.20%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ndp</th>\n",
       "      <td>69.66%</td>\n",
       "      <td>65.33%</td>\n",
       "      <td>67.43%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>People's Party</th>\n",
       "      <td>91.30%</td>\n",
       "      <td>100.00%</td>\n",
       "      <td>95.45%</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Precision   Recall F1-Score\n",
       "Another party (please specify)      97.24%  100.00%   98.60%\n",
       "Bloc Qu<e9>b<e9>cois                92.78%   99.13%   95.85%\n",
       "Conservative Party                  74.81%   67.71%   71.09%\n",
       "Don't know/ Prefer not to answer    58.79%   50.76%   54.48%\n",
       "Green Party                         78.48%   88.75%   83.30%\n",
       "Liberal Party                       66.45%   64.00%   65.20%\n",
       "ndp                                 69.66%   65.33%   67.43%\n",
       "People's Party                      91.30%  100.00%   95.45%"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy = accuracy_score(ground_truth, predicted_classes)\n",
    "precision = precision_score(ground_truth, predicted_classes, average=None)\n",
    "recall = recall_score(ground_truth, predicted_classes, average=None)\n",
    "f1 = f1_score(ground_truth, predicted_classes, average=None)\n",
    "\n",
    "print(f\"Accuracy: {accuracy:.2%}\")\n",
    "\n",
    "df_metrics = pd.DataFrame({\"Precision\": precision, \"Recall\": recall, \"F1-Score\": f1}, index=classes)\n",
    "\n",
    "df_metrics = df_metrics.applymap(lambda x: f\"{x:.2%}\")\n",
    "df_metrics"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "b3ba2566441a7c06988d0923437866b63cedc61552a5af99d1f4fb67d367b25f"
  },
  "kernelspec": {
   "display_name": "Python 3.8.8 ('base')",
   "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": "3.8.8"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}