model validation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "# Анализ тональности текста (Sentiment analysis). Валидация модели\n",
    "\n",
    "В качестве моделей будут рассматриваться рекурентные нейронные сети"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "!pip install -r requirements.txt"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n",
     "is_executing": true
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import warnings\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.metrics import roc_curve, roc_auc_score, confusion_matrix\n",
    "\n",
    "\n",
    "warnings.filterwarnings('ignore')\n",
    "np.random.seed(0)\n",
    "\n",
    "def plot_roc_curve(y_pred, y_true) -> None:\n",
    "    fpr, tpr, _ = roc_curve(y_true, y_pred)\n",
    "    plt.plot(fpr, tpr)\n",
    "    plt.axis([0,1,0,1])\n",
    "    plt.title(\"ROC-кривая\")\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.show()\n",
    "    print(f\"ROC-AUC Score: {roc_auc_score(y_true, y_pred)}\")\n",
    "\n",
    "def plot_confusion_matrix(y_pred, y_true) -> None:\n",
    "    cf_matrix = confusion_matrix(y_true, y_pred)\n",
    "    ax = sns.heatmap(cf_matrix, annot=True, cmap=\"Blues\")\n",
    "    ax.set_title(\"Матрица неточностей\")\n",
    "    ax.set_xlabel(\"Предсказанные значения\")\n",
    "    ax.set_ylabel(\"Действительные значения\")\n",
    "    plt.show()\n",
    "\n",
    "def plot_history_nn(history) -> None:\n",
    "    plt.plot(history.history['accuracy'], label=\"Точность на обучающем наборе\")\n",
    "    plt.plot(history.history['val_accuracy'], label=\"Точность на проверочном наборе\")\n",
    "    plt.xlabel(\"Эпоха обучения\")\n",
    "    plt.ylabel(\"Точность\")\n",
    "    plt.legend()\n",
    "    plt.show()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "outputs": [],
   "source": [
    "train_data = pd.read_csv(\"./data/clear_train_data.csv\")"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "outputs": [
    {
     "data": {
      "text/plain": "                                              comment  toxic\n20  посмотрел комментарии твоим постам процент гов...    1.0\n21                      это утверждал чмоха спалились    1.0\n22  обеих странах партия войны обеих странах эта в...    1.0\n23  причем чувство юмора придумал опроверг писал с...    1.0\n24  агрессия ткнули твое дерьмо обтекай это отрица...    1.0",
      "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>comment</th>\n      <th>toxic</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>20</th>\n      <td>посмотрел комментарии твоим постам процент гов...</td>\n      <td>1.0</td>\n    </tr>\n    <tr>\n      <th>21</th>\n      <td>это утверждал чмоха спалились</td>\n      <td>1.0</td>\n    </tr>\n    <tr>\n      <th>22</th>\n      <td>обеих странах партия войны обеих странах эта в...</td>\n      <td>1.0</td>\n    </tr>\n    <tr>\n      <th>23</th>\n      <td>причем чувство юмора придумал опроверг писал с...</td>\n      <td>1.0</td>\n    </tr>\n    <tr>\n      <th>24</th>\n      <td>агрессия ткнули твое дерьмо обтекай это отрица...</td>\n      <td>1.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.iloc[20:25]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "outputs": [],
   "source": [
    "train_data[\"comment\"] = train_data[\"comment\"].astype(str)  # ошибка в токенизации, что возникает где-то float"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "outputs": [
    {
     "data": {
      "text/plain": "(14412, 2)"
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "outputs": [
    {
     "data": {
      "text/plain": "comment     object\ntoxic      float64\ndtype: object"
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.dtypes"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "outputs": [],
   "source": [
    "X = train_data[\"comment\"]\n",
    "y = train_data[\"toxic\"]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Подготовка данных для обучения"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "outputs": [],
   "source": [
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "outputs": [],
   "source": [
    "# разбиваем текст на токены и создаем их последовательности\n",
    "tokenizer = Tokenizer()\n",
    "tokenizer.fit_on_texts(X)\n",
    "X_clean = pad_sequences(tokenizer.texts_to_sequences(X))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "outputs": [],
   "source": [
    "# разобьем выборку на обучающую и проверочную\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_val, y_train, y_val = train_test_split(X_clean, y, test_size=0.5)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "outputs": [
    {
     "data": {
      "text/plain": "array([[    0,     0,     0, ..., 18290, 12696,  7832],\n       [    0,     0,     0, ...,  6590,  5773,   374],\n       [    0,     0,     0, ..., 19772,   731, 21390],\n       ...,\n       [    0,     0,     0, ..., 62227, 25778, 12904],\n       [    0,     0,     0, ..., 15240,  2553, 21485],\n       [    0,     0,     0, ...,  9086,  3641,  2540]])"
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train[:10]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "outputs": [
    {
     "data": {
      "text/plain": "(7206, 796)"
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "outputs": [
    {
     "data": {
      "text/plain": "(7206,)"
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "outputs": [
    {
     "data": {
      "text/plain": "(7206, 796)"
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_val.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Создание и валидация моделей"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import (\n",
    "    Dense,\n",
    "    Embedding,\n",
    "    LSTM,\n",
    "    GRU,\n",
    "    SimpleRNN,\n",
    ")"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "### Простая рекурентная нейронная сеть"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "outputs": [],
   "source": [
    "max_words = 100000\n",
    "\n",
    "simple_model = Sequential()\n",
    "simple_model.add(Embedding(max_words, 128))\n",
    "simple_model.add(SimpleRNN(32, recurrent_dropout = 0.2))\n",
    "simple_model.add(Dense(1, activation=\"sigmoid\"))\n",
    "simple_model._estimator_type = \"classifier\""
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "outputs": [],
   "source": [
    "simple_model.compile(optimizer='rmsprop',  loss='binary_crossentropy', metrics=['accuracy'])"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding (Embedding)       (None, None, 128)         12800000  \n",
      "                                                                 \n",
      " simple_rnn (SimpleRNN)      (None, 32)                5152      \n",
      "                                                                 \n",
      " dense (Dense)               (None, 1)                 33        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 12,805,185\n",
      "Trainable params: 12,805,185\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "simple_model.summary()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "51/51 [==============================] - 23s 442ms/step - loss: 0.7166 - accuracy: 0.5135 - val_loss: 0.6924 - val_accuracy: 0.5173\n",
      "Epoch 2/10\n",
      "51/51 [==============================] - 23s 444ms/step - loss: 0.6151 - accuracy: 0.6743 - val_loss: 0.6379 - val_accuracy: 0.6616\n",
      "Epoch 3/10\n",
      "51/51 [==============================] - 22s 430ms/step - loss: 0.4874 - accuracy: 0.7978 - val_loss: 0.6006 - val_accuracy: 0.6699\n",
      "Epoch 4/10\n",
      "51/51 [==============================] - 22s 441ms/step - loss: 0.3033 - accuracy: 0.9109 - val_loss: 0.5296 - val_accuracy: 0.7295\n",
      "Epoch 5/10\n",
      "51/51 [==============================] - 22s 435ms/step - loss: 0.1401 - accuracy: 0.9707 - val_loss: 0.5171 - val_accuracy: 0.7476\n",
      "Epoch 6/10\n",
      "51/51 [==============================] - 22s 433ms/step - loss: 0.0614 - accuracy: 0.9867 - val_loss: 0.5284 - val_accuracy: 0.7545\n",
      "Epoch 7/10\n",
      "51/51 [==============================] - 22s 436ms/step - loss: 0.0289 - accuracy: 0.9937 - val_loss: 0.5594 - val_accuracy: 0.7559\n",
      "Epoch 8/10\n",
      "51/51 [==============================] - 22s 436ms/step - loss: 0.0186 - accuracy: 0.9952 - val_loss: 0.6282 - val_accuracy: 0.7420\n",
      "Epoch 9/10\n",
      "51/51 [==============================] - 22s 429ms/step - loss: 0.0103 - accuracy: 0.9975 - val_loss: 0.6744 - val_accuracy: 0.7365\n",
      "Epoch 10/10\n",
      "51/51 [==============================] - 22s 431ms/step - loss: 0.0082 - accuracy: 0.9977 - val_loss: 0.7323 - val_accuracy: 0.7171\n"
     ]
    }
   ],
   "source": [
    "history = simple_model.fit(X_train, y_train, epochs=10, batch_size=128, validation_split=0.1)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history_nn(history)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "outputs": [],
   "source": [
    "y_pred = simple_model.predict(X_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC Score: 0.7542665919529642\n"
     ]
    }
   ],
   "source": [
    "plot_roc_curve(y_pred, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred_int = [1 if pred[0] > 0.5 else 0 for pred in y_pred]  # так как предсказания в интервале 0 и 1, то будем считать, что 1 - больше 0.5, иначе - 0\n",
    "plot_confusion_matrix(y_pred_int, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "### Реккурентная сеть со слоем LSTM"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "outputs": [],
   "source": [
    "model_lstm = Sequential()\n",
    "model_lstm.add(Embedding(max_words, 128))\n",
    "model_lstm.add(LSTM(32, recurrent_dropout = 0.2))\n",
    "model_lstm.add(Dense(1, activation=\"sigmoid\"))\n",
    "\n",
    "model_lstm.compile(optimizer='rmsprop',  loss='binary_crossentropy', metrics=['accuracy'])"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding_1 (Embedding)     (None, None, 128)         12800000  \n",
      "                                                                 \n",
      " lstm (LSTM)                 (None, 32)                20608     \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 1)                 33        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 12,820,641\n",
      "Trainable params: 12,820,641\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model_lstm.summary()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "51/51 [==============================] - 64s 1s/step - loss: 0.6073 - accuracy: 0.6671 - val_loss: 0.6039 - val_accuracy: 0.6338\n",
      "Epoch 2/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.3911 - accuracy: 0.8313 - val_loss: 0.4815 - val_accuracy: 0.7725\n",
      "Epoch 3/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.2085 - accuracy: 0.9479 - val_loss: 0.4100 - val_accuracy: 0.8447\n",
      "Epoch 4/10\n",
      "51/51 [==============================] - 61s 1s/step - loss: 0.1065 - accuracy: 0.9732 - val_loss: 0.4378 - val_accuracy: 0.8460\n",
      "Epoch 5/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.0565 - accuracy: 0.9863 - val_loss: 0.4604 - val_accuracy: 0.8599\n",
      "Epoch 6/10\n",
      "51/51 [==============================] - 61s 1s/step - loss: 0.0326 - accuracy: 0.9911 - val_loss: 0.4810 - val_accuracy: 0.8571\n",
      "Epoch 7/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.0216 - accuracy: 0.9941 - val_loss: 0.5375 - val_accuracy: 0.8419\n",
      "Epoch 8/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.0149 - accuracy: 0.9958 - val_loss: 0.5458 - val_accuracy: 0.8530\n",
      "Epoch 9/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.0119 - accuracy: 0.9965 - val_loss: 0.5836 - val_accuracy: 0.8252\n",
      "Epoch 10/10\n",
      "51/51 [==============================] - 62s 1s/step - loss: 0.0106 - accuracy: 0.9963 - val_loss: 0.5554 - val_accuracy: 0.8294\n"
     ]
    }
   ],
   "source": [
    "history = model_lstm.fit(X_train, y_train, epochs=10, batch_size=128, validation_split=0.1)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history_nn(history)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "outputs": [],
   "source": [
    "y_pred = model_lstm.predict(X_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAESCAYAAAD5d3KwAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAAAovUlEQVR4nO3deXxcdb3/8dfMZGvapPsOdIVPy9JFEEopm8gqSF1+oHi9ghcREVFQUXEDRXADuerlKiKgXlFB5QLqRVDAYlmEUMrS8Cml0NKW7km3NNvM/P44J01a2slkObMk7+fjETJztvnMl+Z8zvme7xJLp9OIiIi0iec7ABERKSxKDCIishslBhER2Y0Sg4iI7EaJQUREdqPEICIiuynJdwAi3WVmaeBFIAmkgUpgK/AJd38m3GYgcA1wFtAcbnc/cK277+xwrI8AFwMDgDLgn8CV7l6fq+8jUih0xyDF7kR3n+Xus93dgN8BPwIwsxLgbwT/zme7+2HAHGAQ8NdwPWZ2FXAhMN/dZwEzgRaCBCLS7+iOQfqM8ER/ALA5XPT/gLi7X9G2jbs3mNlngEXAe8zsL8BVBIljXbhNi5l9Plxf5u7Ne3zO1cAI4FPAHwnuML4X/qwGJgM7gfPdvdbM7gBedPfvm9kQoBb4qbtf3eGuJ0Zwt/JRd/+Hmc0BvguUA2OBh9z9P8LP/3b43bYDFUC5u0/shSIUAXTHIMXvETNbbGZrgKXhsgvC33OBBXvu4O5p4O/APGAa0ODur+yxTYO7/3rPpLCHjwKN7v6u8P3bgBvcfQZwO/Crvezzvb0sO9HdDwW+CXwjXPZp4GvufhRwMPBuMzvczA4NP/dQd58JfChDfCLdosQgxe7E8AT5LoJnDI+7+/oO60v3sV85wfOGFN37Ozif4CR+eYdli939sfD1bcBsMxvettLMjgWmEtxl7M1ooD58/RFgSFjNdTPBdxtE8AylBBjZjZhFsqLEIH2Cuy8iOEnfamYTw8ULgePMbLd/5+H744DHgSVAqZlN3WObCjP7i5mNC38/F/68O9zkDoKT/40ddmvt8DoW/iTD92XADwkecO85QNkjZvYiQaL5v3DZY8AZwMvh8lVAzN1XApcCC81sNXBnp4Uj0kVKDNJnuPtvgCeAm8JFvwd2ADeZ2QCA8PePCOrn73H3JuA7wG1mNjrcphz4ATDQ3de4+xnhA+5Z7n5fh4/8JnC0mZ0Uvp9lZjPC1xcBCzu0avpk+Hm+l9DbqpIOBn4cxnEE8AV3/yMwnuBOIxFuPxLYCBwCnNflghLphBKD9DWXAqeb2anu3gqcQpAEasKr8mfD9ye7ewuAu18H/IGgpdJzwGKCq/2zM31Q+PzhywTPDeLAWuBbZvYCMB/4cIfNtwDf3sehHjGzxcADwM3hQ/DrgWfN7BngSwR3P1PN7B3hZ75HTWklKjENuy3Sc2Z2AvDj8MpfpKjpjkFERHajOwYREdlNZHcMZnaUmT26l+VnmdnTZvaEmX0sqs8XEZHuiSQxmNmVwK0EvTI7Li8laO1xCnA8cFFbSxARESkMUQ2J8SrwXt7a83M6sMzd6wDM7J8E7cnvznSwmpqadDyuxyEAqVQKlUVAZdFOZdGuO2WRTqd3dS5Jp4OOJuk0JFNvXZ5Kp0mlg99NrWnisfaOKenwP62pYJtYrH3ftvUtyWBdLkwdVrrx8MMP73JnyEgSg7v/oUMno46qCZrttdkGDO7sePF4nNmzZ/dSdMWttraW6dOn5zuMgqCyaKeygB1NrWzc3sRji15m7PhxtCTTtKZStCbTbNrRzPbGVlbVNfD065tJxGO0ptK0tKZYs6Wxx589fGAZiXiMkniMeDxGPBajbkczB44exMDyEhLxGIlYLPgdj7GtsZWpowZxyLjqXcvisbbfEA8zSlVFKVUVJcTCZbt+A7E93retD36C46xZtmRFd75PrgfR2wpUdXhfRfsQACJS5NLpNMlUmtbwJ5lMs7WxhfqGFpqTSVZubmBNfSMbtjURj8VIplIk02mSKUil0iTTaV5eu5XqilJa207sqTS+dhtlJXHisVhwxR5um0oR7r/nJfjavcYXiwVX72OqK5gzeRiliTgliTj1Dc0cOn4wpYkYiXg8/B0jmUpzwLBKShPxXSf+0kSc0dUVVJTGKS9JMKAssdfPKgRrurlfrhNDLXCgmQ0j6GR0HPD9HMcgIhms2LSDlZsbaG5N0ZJM8cq67ZQk4rQkU6yqa2BbYyuvb2qgNZkiDbQmg5P3qrqdnR67o7KSOANKEx2uliERixGLxVi+YQcz9hvMwNLgavvYA0eyeUcTh40fvOuKfM/9UmkYPKCU1m2beNv0yQyqKAlO/PEYA8oSjKqqIBGPRVNofUxOEoOZnQcMcvdbzOwK4K8ED75vc/fVuYhBpL9IptI0tiR5c8vO4Mo92X4Vn0qnedTX07GV+oJXNpBKwasbttPUmsrqM6aOGsS2xhbedsDQXSffuVNi1DW0MGP8YBKJ4Oo6EQ/WlSRi7D+0krKSOCMGlbHf0EoqSqO50q6tbWL65OGdbyj7FFlicPfXCSZFwd3v7LD8fjQBiki3NLYkWbB0A8+v2kJLKkVjc5JnV9ZTkmom9fdNLF61pfODhEoTwdVzSzLNwLIEM/YbzAHDBjJ4QCmzDxjChOFBFUppIsa4IQMoL0noiruf0EQ9IgUqnU6z6I163tjcQM2KOn75xFufI1aVlxCPxyghhY2rZN7UERwyrpqhA8tobk1x0OhBxGPBFXsiHicRi1FeGmf2/kMoSagVk+ydEoNIjjS2JHlpzRaaW9O0JIP6+6XrtlNRGieZStOSTLNycwNL1mxh7dZG1m1tessxRgwq45ITpnK8jWTCsMpdJ3e1SpLepMQg0k1tdfktyRTNyRSvbdjBxu3NLFpZx+ubGojFoCWZYntjKy+s3pJ1/T3AyKpyTrSRHDJuMMfbSPYfWsno6nJiMVXlSPSUGEQ60ZJM0dCcpG5HM8+urOOp5Zt56c0tvLh6a6f7zthvMGWJOIeOH8zIQeVMGTWQY6aOoCxsJlmaiDF28ABKEjFK43FKwoe2SgCST0oM0q9tbWzh6dc2s3Tddppak/zrtc2UJOI8v6qeQeUlvLmlcS9t5AOTRw7kjEPHMnxQ2a6WOeOHDmC/oZVMGjEwx99EpPcoMUi/k0qleX3TDj5792IWrazfbV1JPEZlWYL9h1UyeEApxx44gu1NSQ4eW82gihKqK0o4espwRlVV7P3gIn2AEoP0SdubWqlvaGZV3U4efnk9Ty3fxLamVup2NFPX0LJru8qyBB89ZhLzZ49n3JAKKsv0JyGivwIpeuvDFjwrNu/grmdW8er67ayuf2sv3MEDSjn+oJHEYjBh+EDmTB7G3Ckj8hCxSGFTYpCi0NiS5PlVW9jR3Ertm1u5d9EaGpsaWfGL5Xvd/uxZ4zhs/GBGVVcwprqCwycMVecskSwpMUjBSKXSPLeqni07W1iyZivLN+zgtY3bWbm5gY3bm9+yfXlJjPmzxtHQnOSdB49m3OABjBlczuQRg4grCYh0mxKDFISdzUkOvfqve20BdMSEocyfNZ6qilLmTh3OwLISJo6oZMWrr6hTl0gElBgkL9LpNG9s3slPFrzKoy+v321M/D9eMpfKsgSTRgykvKRwhzQW6auUGCRnkqk0i1bW8bfa9fzsseW73R0cOXEYZ84cy4fnTFDnLpE8U2KQSC3fsJ3fPf0Gd9esYvOO9ucEZYk4Mw4YzKfeMZW5U0ZENgSziHSdEoP0qmXrt3Pfc6t56rXNPPXa5t3WHTy2mtMPHcMxB45g9v5DdGcgUqCUGKTbkqk03/zTEh57ZQOtqTQrNjXstn7EoHLedsAQPnjUAcybOoJSDfMsUhSUGCRrb27Zyd+WrKO+oYWFr27kyeXtdwRHTRrGUZOGUdfQwtmzxnGCjWJQuf55iRQj/eVKRq3JFPcsWs23/lJLfYehJABGVZVz6iFjuOqM6QU9IbqIdI0Sg7xFc2uKB15ay++eXsnCZZt2W/eDc2fyjmmjGTygNE/RiUjUlBiEVCrNMyvqqFlRx79e28QjvmHXulFV5bx90jC++74ZDFTVkEi/oL/0fu73Nav43N2L37L80hOn8v7D92Oi5hUQ6XeUGPqhdDrNtX+u5c6nVrKzJQmAja7ipg/MCnsbx9WUVKQfU2LoR1qSKVbV7eTSO5/lpTXBtJRzpwzn62cdgo2pynN0IlIolBj6iVQqzaxrHmRHc3LXspe/eZp6HIvIWygx9HHpdJrbFr7ON/+0BIChlaXceM4s5h2oDmcisndKDH1UazLFg0vW8YXfP8+2plYgqDb69YVH6fmBiGSkxNDHtCRT/OChpdz86Ku7lp1y8Gi+f85MqivU90BEOqfE0EfsaGrlq/e+yH3PraE1HM76XTPGct38wxhcqYQgItlTYugDFizdwL/f9q9d7/9j3iSuOPkgdUgTkW7RmaOI3figc8tjy2lsSQHw8eMm84XTpmm+YxHpESWGInXHwtf44cPLgOAZwuUnH8T0sdV5jkpE+gIlhiKzor6ZC7/9MKvrdwLw4OXHcdBodU4Tkd6jxFAklq7bxmfvWswLq7fsWvbny+YpKYhIr1NiKAL3L17Dp36zCIDq8jjfP2c2Jx88Wv0RRCQSSgwF7hv3L+G2ha8Fr88+hLcP2cn06WPyHJWI9GWRJAYziwM3AzOBJuBCd1/WYf1ngfOAFHCdu98TRRzFbP3WRuZc/3fCLgncccHbOcFGUVtbm9/ARKTPi+qOYT5Q4e5Hm9kc4AbgbAAzGwJ8GpgKDASeA5QYOrjqnhe486mVAEwYXskvP3okE4ZrXgQRyY2oEsM84AEAd3/SzI7osG4HsIIgKQwkuGsQggHvzvjhP6l9MxgS+71vG8+N58zKb1Ai0u9ElRiqgS0d3ifNrMTdW8P3bwBLgARwfWcHS6VS/aIK5Z4lW3YlhXv/bRJlidhbvndjY2O/KItsqCzaqSzaqSx6LqrEsBXo2I4y3iEpnA6MBSaF7/9qZgvd/V/sQzweZ/r06dFEWiCWrNnKLU8vB2DRV09m6MCyvW5XW1vb58siWyqLdiqLdiqLdjU1Nd3aL6oB+RcCZwCEzxhe6LCuDtgJNLl7I1APDIkojqLwu6dXcsYPHwPgpnNn7TMpiIjkQlR3DPcAJ5vZ40AMuMDMrgCWuft9ZvZO4EkzSwH/BB6KKI6Clk6n+eafanc1R/3MOw9k/uzxeY5KRPq7SBKDu6eAi/dY/HKH9V8Hvh7FZxeLdDrNaTc9hq/bBsCdFx7F3Kkj8hyViIg6uOXFuq2NnPC9R9nZEsy/vPy6MzQiqogUDCWGHNre1Mpnfvscf6tdt2tZzVfeqaQgIgVFiSGHrvrjC7uSwpWnGZecMDXPEYmIvJUSQ47cvvA17lu8BoBXvnU6pYmoGoSJiPSMzk45ULNiM9fcvwSA+y+dp6QgIgVNdwwRenXDdj5861Os2dIIwBdPn8Zh+w3Oc1QiIpkpMUSkobmVk274BwDTxlTxtbMOZu4UNUcVkcKnxBCRH4XzMX/wyP25/r0z8hyNiEj2lBgicOXvF3PXM6sA+OqZB+c5GhGRrlFi6GWfv3sxd9cESeG+S4+hskxFLCLFRWetXnTtn5bsSgoPXn4cB42u6mQPEZHCo8TQSx7x9dz6z2AwvIc/ezyTRw7Kc0QiIt2jBvW94OnXN3PB7U8D8PlTTUlBRIqa7hh6wS0Lggl2bjp3lobNFpGipzuGHnpjcwMPLQnGPzp71rg8RyMi0nNKDD2QTqc59ruPAPC1Mw8mFtMoqSJS/JQYeuDHYSe2oZWlfHTepE62FhEpDkoM3fTEq5u44aGlADz+xZPyHI2ISO9RYuiGLQ0tfPBnTwJw+/lvZ0BZIs8RiYj0HiWGLmpsSXLafy4A4KyZ4zhx2qg8RyQi0rvUXLULlq7bxik/CJJCVUUJN507K78BiYhEoNPEYGZVwBeAccCfgOfdfVnUgRWauh3Nu5LCsQeO4Ff/cVSeIxIRiUY2VUm3AcuBA4G1wM8jjahAXfjLZwD4wmnTlBREpE/LJjEMd/fbgBZ3fzzLffqUT/1mETUr6pg8ciCfOGFKvsMREYlUVid5M5sW/t4PaI00ogKTTqe5f/EaAH750SPzHI2ISPSyefh8GXA7MB34PfCJSCMqMPc//yYAJ00bxX5DK/McjYhI9LJJDBPd/ei2N2Z2DrAoupAKx6KVdVz2m+CrXv3uQ/IcjYhIbuwzMZjZmcAxwAfNbG64OA6cDdyVg9jyKp1Oc+5Pg05s173nMPYfprsFEekfMt0xLAaGAzsBD5elgN9GHVQheGL5JpqTKY6cNIzzjjog3+GIiOTMPhODu78B/MLMfuXuqbblZjY2J5Hl2Xk/ewoI7hZERPqTbJ4xXG1mnwDKgEpgKdCnK9xvC6foBJg6SrOxiUj/kk1z1XcD+wG/JmiZtDrSiArAzY8GHbv/dZVGTRWR/iebxPCmuzcBVeFQGGURx5RXNz60lI3bm3nv28Yzqroi3+GIiORcNolhlZl9FNhhZtcDQ6INKX9SqTQ//PsrAHznfTPyHI2ISH5kkxg+Dvwd+DywBvhgpBHl0YJXNgDwrhljKU30u5E/RESAzP0YSgieL9S5+yPhsruB/wTOzU14uVOzYjPn3/40ABcdOznP0YiI5E+mVkm/JhgXaayZHQK8RjCy6n92dlAziwM3AzOBJuDCjkN1m9npwNeBGFADfNLd0939Ej21ZWcL7/vvJwC46LjJzNx/SL5CERHJu0z1JVPc/UPAaQTVSdcAJ7r79Vkcdz5QEQ6l8UXghrYV4fwO3wPOdPejgNeBEd2Kvhe8tGYLM695EIDz507kqjOm5ysUEZGCkOmOYSuAuzeHdwCnuPvmLI87D3gg3P9JMzuiw7q5wAvADWY2GbjV3TdkOlgqlaK2tjbLj85eczLF2f/zOgAnTBrEuQfGI/mc3tTY2FjwMeaKyqKdyqKdyqLnsp3ac10XkgJANbClw/ukmZW4eyvB3cGJwCxgO/CYmT3h7kv3dbB4PM706b1/JX/N/S8B8LFjJ/Hldx3c68ePQm1tbSRlUYxUFu1UFu1UFu1qamq6tV+mxHCImd1J8Byg7TUA7n5eJ8fdClR1eB8PkwLAJuBpd18LYGYLCJLEPhNDFNLpNLcvfB2Az586LZcfLSJS0DIlhnM6vP5JF4+7EDgLuMvM5hBUHbV5FjjUzEYA9cAc4GddPH6PfSRsgfTe2eMpK1HTVBGRNpkG0ftHD457D3CymT1OcMdxgZldASxz9/vM7EvAX8Nt73L3F3vwWV3W2JJkwdLgscb179MgeSIiHWX7jKFLwtFYL95j8csd1v+WPA7ffeNDQa3Vl06fRnlJIl9hiIgUpH5Xh/LcG/XcsmA5A0oTXHDMpHyHIyJScDq9YzCz8cB3gFHA3cDz7v5U1IFFIZlKM/+/FgJw/XsP07MFEZG9yObMeAtwG1AKLCCLns+F6scP7+p8zfzZ4/MYiYhI4comMQxw94eBtLs70BhxTJFpm2dh8ddOyXMkIiKFK5vE0GhmpwKJsOlpUSaG+oZmmlpTHDlxGIMrS/MdjohIwcomMVwEXEDQY/lzwCcijSgi//3oqwC8//D98hyJiEhhy6a56vuAT7h7XdTBROmnC5YDcObMsXmORESksGVzx1AC/M3Mfm1mJ0QcTyQeeXk9AEdOHEZlWSRdN0RE+oxOE4O73+DuhwM3AZeYWU7HNOoNX7036Fh9wzkz8xyJiEjhy6YfwwCC6qSPEAxv8fWog+pta+p3EovB/sMq8x2KiEjBy6Ze5Xng9wTPGZZ1tnGh2byjmVQazp41Lt+hiIgUhYxzPodDZc8GmsNlZRBM3pOb8Hru4796BoAjJg7LcyQiIsUh0x3DL4HzCIbMThNUIxG+nhxxXL0inU7z9OtBY6pzj9g/z9GIiBSHTMNut03Gc467P922vJhaJv3h2dUAnHLwaI2LJCKSpUxVSfOAQ4DLzezGcHEcuBQ4NAex9dgDL64F4IpTDspzJCIixSNTVVI9MAYoB9p6haWAKyOOqdf8rXYdANPGVOc5EhGR4pGpKulF4EUzu8Xd38xhTL1i0crg2cJJ00blORIRkeKSqSrp9+7+fuBZM0uHi2MEo6wWfNvPBUs3AnDxCVPyHImISHHJdMfw/vB3UQ4udMfjr7H/sAEcMWFovkMRESkq2fR8fme4XRz4EfBVd78z6sB6Ytn6bdQ1tPD2icOIxWKd7yAiIrtk04bzW8ArwGXAMcDFkUbUC77whxcA+OBRB+Q5EhGR4pNNYmgA1gGt7r6WoINbQatZUUdlWYITTQ+eRUS6KpvEsBV4ALjLzD4JrI82pJ7ZsK0JgGOmjshzJCIixSmbQfTOAaa4+xIzOwS4NeKYeqRtbCTN1CYi0j3Z3DGMBK4xs5eAa2nv7FZw/vLCmzy7sp5JIwZy6iFj8h2OiEhRyiYx/Az4FcGD518AP480oh74xv1LAPjBubPyG4iISBHLpiqpwt3vC1//r5ldEWVAPdGaSjGyqpxZ+w/JdygiIkUrqzmfzewwgPB3wbZK2tbYyjvUEklEpEeyuWO4DLjNzMYCa4CPRRtS97yxuYGm1hSV5Yl8hyIiUtQyJgYzqwbc3d+eo3i67fFXg7GRjp48PM+RiIgUt31WJZnZpcBiYLGZnZq7kLrnqdc2AzBnihKDiEhPZHrGcB5gwNHAZ3ISTQ+0JoNHH9UVpXmORESkuGVKDI3u3uzuG4GyXAXUXfctXsPgAUoKIiI9le1EyAU9ROn/Llqd7xBERPqMTA+fDzGzOwmSQttrANz9vMgj64L7F68B4B+fPyG/gYiI9AGZEsM5HV7/pCsHNbM4cDMwE2gCLnT3ZXvZ5s/Ave7epePvac2WRsZUVzCksuBrvERECl6mGdz+0YPjzifoMX20mc0BbgDO3mOba4EeT6/W2JKk9s2tvO9tGjRPRKQ3ZPuMoavmEQzVjbs/CRzRcaWZvR9ItW3TE2vqdwIwYXhlTw8lIiJk1/O5O6qBLR3eJ82sxN1bzexQgqaw7we+ls3BUqkUtbW1e133y0VB/4WK5vp9btOXNDY29ovvmQ2VRTuVRTuVRc9lM+fzeOA7wCjgbuB5d3+qk922AlUd3sfdvTV8/e/AeOBhYCLQbGavu/s+7x7i8TjTp0/f67rkS4uBev7tpNlUlkWV5wpHbW3tPsuiv1FZtFNZtFNZtKupqenWftmcSW8heEbwVWABwdDbczrZZyFwFsGsb3OAF9pWuPuVba/N7Gpgbaak0JnHX93EiEHl/SIpiIjkQjbPGAa4+8NA2t0daMxin3uARjN7HPgBcLmZXWFm7+5BrHu1qm4n1RVKCiIivSWbM2pjOFZSIrz67zQxuHsKuHiPxS/vZburswlyX7bsbAFgkBKDiEivyeaO4SLgAmAE8DngE5FG1AUPvrQWgFMOHp3nSERE+o5OL7XdfRXwgRzE0mUt4cB5ZxxWsNNQi4gUnWxaJb1JMGtbDBgGLHf3gnjkv2LzDgDGDK7IcyQiIn1HNncMuy7HzWwCcHWUAXVF7ZvbGFpZqhZJIiK9qEs9n919BTAtoli6bMHSDTQ0J/MdhohIn5JNVdJvCKqSAMYC6yKNKEttLZKq1CJJRKRXZXNW/R1QF75uBJ6JLpzsvbQmGHHjspMOzHMkIiJ9SzaJ4XPuPi/ySLpo3dagO8Wh4wfnORIRkb4lm8Sw2cw+DTjBiKi4+4ORRpWFf/gGAIZoOk8RkV6VTWLYBMwKfyB43pD3xLB4VVCVdMAwDbctItKb9pkYzOx37n6uu1+Qy4Cy0ZJM8drGHbxrxlhKElFNKSEi0j9lOquOzFkUXbRycwMANrqqky1FRKSrMlUlTTGz6/a2wt2viiierDS2BH0XDho9KJ9hiIj0SZkSQwPBA+eCs7puZ75DEBHpszIlhrXu/oucRdIFL6/dBsABwwbmORIRkb4n0zOG7s0JlwOLVgb97SaPVGIQEelt+0wM7v65XAbSFSs2NTCwLEFFaSLfoYiI9DlF19YznU6zfOMODlSLJBGRSBRdYtje1ArAiEHleY5ERKRvKrrE0DZr27ypw/MciYhI31R0iaFtuG31eBYRiUbRnV23NQaJoTQRy3MkIiJ9U9ElhhdWB4PnjR08IM+RiIj0TUWXGFKp4BnDgRoOQ0QkEkWXGNp6PQ8q15SeIiJRKLrEsG5rEwBVFZqgR0QkCkWXGOJ65iwiEqmiSwyP+gZm7qd5nkVEolJ0iaE5mWJnOB+DiIj0vqJKDG0T9Bw9Wb2eRUSiUlSJob4h6Nw2qroiz5GIiPRdRZUYNu0IWiSNUWIQEYlMUSWGZ14PJuiZpAl6REQiU1SJoW2cpAnDKvMciYhI31VUieGhJeuoKi9haGVZvkMREemziioxAAwsLyGuXm4iIpGJZMAhM4sDNwMzgSbgQndf1mH95cAHwrd/cfdrsjluSzLNoeOreztcERHpIKo7hvlAhbsfDXwRuKFthZlNBj4EzAXmAKeY2YxsDrrkza0kdLcgIhKpqIYonQc8AODuT5rZER3WvQGc5u5JADMrBRozHSyVSlFbW8ugsjib6rdSW1sbUdiFr7GxsV9//45UFu1UFu1UFj0XVWKoBrZ0eJ80sxJ3b3X3FmCjmcWA7wGL3H1ppoPF43GmT58OsZXMmDgmeN1P1dbW9uvv35HKop3Kop3Kol1NTU239ouqKmkrUNXxc9y9te2NmVUAvw63uSTbg25vatWUniIiEYsqMSwEzgAwsznAC20rwjuFe4HF7v7xtiqlzrT1YdjR3NrJliIi0hNRVSXdA5xsZo8DMeACM7sCWAYkgOOBcjM7Pdz+S+7+RKYDrt8WDIdhY9QqSUQkSpEkBndPARfvsfjlDq+7PNjRpu3NAAwsS3Q/MBER6VTRdHB7aU3wLHu/oRoOQ0QkSkWTGNoeOU8dNSivcYiI9HVFkxh83TYAykuKJmQRkaJUNGfZ/3txLRCMlSQiItEpmsTQ3Jpi2piqzjcUEZEeKZrE0JpMc4KNyncYIiJ9XtEkhpZUSr2eRURyoCgSQzqdJp2GknhRhCsiUtSK4kzbmgp+NyezGj1DRER6oCgSQ3MyDcCEYQPzHImISN9XFImhNRUkhimjlBhERKJWFImhzdjBA/IdgohIn1cUiaHtjqE0URThiogUtaI406aDvEBVhXo9i4hErSgSA0BpIkZFqYbcFhGJWlEkhuZkmkRcndtERHKhKBJDLAaNLal8hyEi0i8URWIAmDJSTVVFRHKhKBKDhsMQEcmdojjbtqb0jEFEJFeKIjGkgU07mvIdhohIv1AUiSEGTBtTne8wRET6haJIDK2pNAPUh0FEJCeKIjEkU7CjuTXfYYiI9AtFkRjSwMThaq4qIpILRZEYAEo0raeISE4UTWKYPHJQvkMQEekXiiYxlKofg4hIThRNYginZBARkYgVTWIYO6Qi3yGIiPQLRZMYEjFVJYmI5ELRJIYSPWMQEcmJokkMcSUGEZGcKJrEMKSyNN8hiIj0C0WTGA4aVZXvEERE+oWSKA5qZnHgZmAm0ARc6O7LOqz/GPBxoBW41t3/1NkxVZUkIpIbUd0xzAcq3P1o4IvADW0rzGwMcBlwDHAqcL2ZlWc6WJkGVhURyZmoEsM84AEAd38SOKLDuiOBhe7e5O5bgGXAjEwHi6mpqohIzkRSlQRUA1s6vE+aWYm7t+5l3TZgcKaD7V9dsrGmpmZF74dZnGpqavIdQsFQWbRTWbRTWewyoTs7RZUYtgIdnxbHw6Swt3VVQH2mgx1++OEjezU6ERHZp6iqkhYCZwCY2RzghQ7r/gUca2YVZjYYmA68GFEcIiLSRbF0uvdHp+vQKmkGwZTNFxAkimXufl/YKukigsR0nbv/odeDEBGRbokkMYiISPEqmg5uIiKSG0oMIiKyGyUGERHZTVTNVbsliqE0ilEW5XA58IHw7V/c/ZrcR5kbnZVFh23+DNzr7j/JfZS5kcW/i9OBrxM0+KgBPunuffIhYhZl8VngPCBF0MDlnrwEmkNmdhTwHXc/YY/lZwFfIzhv3ubuP+vsWIV2xzCfXhxKo4jNZ9/lMBn4EDAXmAOcYmYZe44Xufnsoyw6uBYYmsug8mQ++/53UQV8DzjT3Y8CXgdG5CHGXJnPvstiCPBp4GjgFOCm3IeXW2Z2JXArULHH8lLgBwTlcDxwkZmN7ux4hZYYenUojSKWqRzeAE5z92R4NVgKNOY+xJzJVBaY2fsJrgofyH1oOZepLOYS9Be6wcweA9a5+4bch5gzmcpiB7ACGBj+pHIeXe69Crx3L8unE3QTqHP3ZuCfwHGdHazQEsNeh9LYx7pOh9IoYvssB3dvcfeNZhYzs+8Di9x9aV6izI19loWZHUpQXfC1fASWB5n+PkYAJwJfAE4HPmNmB+U4vlzKVBYQXEAtAZ4FfpjLwPIh7AvWspdV3TpvFlpi6NWhNIpYpnLAzCqAX4fbXJLj2HItU1n8OzAeeBg4H7jCzE7LbXg5laksNgFPu/tad98OLABm5Ti+XMpUFqcDY4FJwAHAfDM7MsfxFYpunTcLLTFoKI3APsvBzGLAvcBid/+4uyfzE2LO7LMs3P1Kdz8qfNh2B3Cju/flKqVMfx/PAoea2YjwynkOwRVzX5WpLOqAnUCTuzcSnAiH5Di+QlELHGhmw8ysjKAa6YnOdiqoVknAPcDJZvY44VAaZnYF7UNp/BB4jCChfTn8n94X7bMcgATBQ6TysBUKwJfcvdP/2UUq47+J/IaWc539fXwJ+Gu47V3u3lcvnKDzsngn8KSZpQjq1R/KY6w5Z2bnAYPc/ZawXP5KcN68zd1Xd7a/hsQQEZHdFFpVkoiI5JkSg4iI7EaJQUREdqPEICIiu1FiEBGR3RRac1URzGwi8DxB2/w2D7v7N/ax/R3Ab7vbh8HMXgdWAkmCi6VNwEfcfVsXjvFFgo52zwP/5u63mtn5wObuNqvtEFeKoJnyIOBj7v5Mhn0udfcfd+fzRNooMUihWrLnKJERO6WtX4yZfYdgOtqsh1Jw92+H+04ELgRudfc7ejmuU4GrgTMzbP8VQIlBekSJQYqGmSWAnwL7Ewx5cJ+7f6XD+oOA2wmGF44D57n7G2Z2PXAswVX3je5+d4bPiBH0kvVwZMrbgckd9v2dmV0CfITgSv5pd7+s7a4FeB9wsJl9LYxhLXAQQU/1X4SjBP/Z3Q/vSlyhCQS9etsGD/wkwSCKaeA9BEPSDzOzmwlGF/0JcGAYx1fc/dFOji8C6BmDFK6DzezRDj/jCRLCk+5+KsFouxfvsc/JBEOnvJNgXoLBYe/wSe4+j2CQuS+HwzLv6UEzewT4O8HJ95cEJ9oN7j43POa1ZjaC4G7i0nDI59o9Bm/7FsHdTsdqr1sJEgnAh4HbuxjXv8xsVfidPxcuPwh4V7j/EuBUd/8WQdXVJQR3LRvd/TjgbOC/9nJskb3SHYMUqrdUJZlZNfB2MzuRYHCwPefj+DnB6KIPEIwoeRVwGHC4mT0ablMKTASe22PfXVU2HT5vOvA3AHffZmZLgCkEieFzZjaJYNyZWKYv4u5LzKzEzCYA5xIkmYu6EpeZXUcwKNz6cPl64Bdmth2YxlvHvzmMYGyxo8L3JWY2wt03ZopVBHTHIMXlfKDe3T9EMDFLZVj10+Zs4DF3Pwm4myBJvAw8EiaZdwB3EYxdn41agqqetolwDgNeAz4GXOzuxwOzCeZCaJNi739XPwe+S5Dw6rsR11eAccAl4SCS1xDM4nchwYBxbeXQ9vtl4Dfh8U8nKI/N2X1t6e+UGKSY/B04zcwWAP8NvEJwsmzzDPANM3uYoJrpR8D9wPZw8poaIN2F1ka3AMPN7J/Ao8A17r6eYCTPx8LPWQ881WGf9UBZ+AC7o7sJZh68NXzfpbjcPUWQBL5C0DppIcFdwmMEiaGtHJaY2f8QPIuZZmb/AB4HVoTHEOmUBtETEZHd6I5BRER2o8QgIiK7UWIQEZHdKDGIiMhulBhERGQ3SgwiIrIbJQYREdnN/wf/Aq1QO1EwGAAAAABJRU5ErkJggg==\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC Score: 0.8877520725678216\n"
     ]
    }
   ],
   "source": [
    "plot_roc_curve(y_pred, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred_int = [1 if pred[0] > 0.5 else 0 for pred in y_pred]  # так как предсказания в интервале 0 и 1, то будем считать, что 1 - больше 0.5, иначе - 0\n",
    "plot_confusion_matrix(y_pred_int, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "### Реккурентная сеть со слоем GRU"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "outputs": [],
   "source": [
    "model_gru = Sequential()\n",
    "model_gru.add(Embedding(max_words, 128))\n",
    "model_gru.add(GRU(32, recurrent_dropout = 0.2))\n",
    "model_gru.add(Dense(1, activation=\"sigmoid\"))\n",
    "\n",
    "model_gru.compile(optimizer='rmsprop',  loss='binary_crossentropy', metrics=['accuracy'])"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " embedding_2 (Embedding)     (None, None, 128)         12800000  \n",
      "                                                                 \n",
      " gru (GRU)                   (None, 32)                15552     \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 1)                 33        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 12,815,585\n",
      "Trainable params: 12,815,585\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model_gru.summary()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "51/51 [==============================] - 48s 904ms/step - loss: 0.6020 - accuracy: 0.6768 - val_loss: 0.5422 - val_accuracy: 0.7393\n",
      "Epoch 2/10\n",
      "51/51 [==============================] - 47s 918ms/step - loss: 0.3169 - accuracy: 0.8958 - val_loss: 0.4446 - val_accuracy: 0.8280\n",
      "Epoch 3/10\n",
      "51/51 [==============================] - 46s 911ms/step - loss: 0.1515 - accuracy: 0.9611 - val_loss: 0.4144 - val_accuracy: 0.8294\n",
      "Epoch 4/10\n",
      "51/51 [==============================] - 47s 919ms/step - loss: 0.0748 - accuracy: 0.9807 - val_loss: 0.4613 - val_accuracy: 0.8447\n",
      "Epoch 5/10\n",
      "51/51 [==============================] - 47s 914ms/step - loss: 0.0369 - accuracy: 0.9900 - val_loss: 0.5067 - val_accuracy: 0.8419\n",
      "Epoch 6/10\n",
      "51/51 [==============================] - 47s 912ms/step - loss: 0.0212 - accuracy: 0.9941 - val_loss: 0.9164 - val_accuracy: 0.8114\n",
      "Epoch 7/10\n",
      "51/51 [==============================] - 46s 911ms/step - loss: 0.0140 - accuracy: 0.9955 - val_loss: 0.6393 - val_accuracy: 0.8197\n",
      "Epoch 8/10\n",
      "51/51 [==============================] - 47s 913ms/step - loss: 0.0113 - accuracy: 0.9961 - val_loss: 0.6535 - val_accuracy: 0.7989\n",
      "Epoch 9/10\n",
      "51/51 [==============================] - 46s 905ms/step - loss: 0.0094 - accuracy: 0.9960 - val_loss: 0.6910 - val_accuracy: 0.7961\n",
      "Epoch 10/10\n",
      "51/51 [==============================] - 47s 920ms/step - loss: 0.0074 - accuracy: 0.9968 - val_loss: 0.7276 - val_accuracy: 0.7989\n"
     ]
    }
   ],
   "source": [
    "history = model_gru.fit(X_train, y_train, epochs=10, batch_size=128, validation_split=0.1)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history_nn(history)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "outputs": [],
   "source": [
    "y_pred = model_gru.predict(X_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC Score: 0.8733578308404863\n"
     ]
    }
   ],
   "source": [
    "plot_roc_curve(y_pred, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 2 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred_int = [1 if pred[0] > 0.5 else 0 for pred in y_pred]  # так как предсказания в интервале 0 и 1, то будем считать, что 1 - больше 0.5, иначе - 0\n",
    "plot_confusion_matrix(y_pred_int, y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Примеры"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "outputs": [],
   "source": [
    "sample_data = train_data.sample(10)\n",
    "X_sample = sample_data[\"comment\"]\n",
    "X_sample = pad_sequences(tokenizer.texts_to_sequences(X_sample))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "outputs": [],
   "source": [
    "y_simple_rnn = simple_model.predict(X_sample)\n",
    "y_gru = model_gru.predict(X_sample)\n",
    "y_lstm = model_lstm.predict(X_sample)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "outputs": [],
   "source": [
    "data = {\n",
    "    \"comment\": sample_data[\"comment\"].to_list(),\n",
    "    \"true_prediction\": sample_data[\"toxic\"].to_list(),\n",
    "    \"simple_rnn_prediction\": [1  if value[0] > 0.5 else 0 for value in y_simple_rnn],\n",
    "    \"gru_prediction\": [1  if value[0] > 0.5 else 0 for value in y_gru],\n",
    "    \"lstm_prediction\": [1  if value[0] > 0.5 else 0 for value in y_lstm],\n",
    "}"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "outputs": [
    {
     "data": {
      "text/plain": "                                             comment  true_prediction  \\\n0  статье говорится хорош управлении вс упр рлось...              0.0   \n1  это правила работы документами сотрудник захоч...              0.0   \n2                       почему бцж появляется шрамик              0.0   \n3    посмотрите одета это любому якутске происходило              0.0   \n4                               ивашка простоквашкой              1.0   \n5  это эрефия метода сделать говно запретить бону...              1.0   \n6  честно говоря базовые потребности полностью по...              0.0   \n7  жалко р ближайшие несколько лет кредиты брать ...              0.0   \n8  выявлен каким анализам случайно рентгену ежего...              0.0   \n9  говорит регионы развиваются работа достойная з...              0.0   \n\n   simple_rnn_prediction  gru_prediction  lstm_prediction  \n0                      0               0                0  \n1                      0               0                0  \n2                      0               0                0  \n3                      0               0                0  \n4                      0               1                1  \n5                      1               1                1  \n6                      0               0                0  \n7                      0               0                0  \n8                      0               0                0  \n9                      0               0                0  ",
      "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>comment</th>\n      <th>true_prediction</th>\n      <th>simple_rnn_prediction</th>\n      <th>gru_prediction</th>\n      <th>lstm_prediction</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>статье говорится хорош управлении вс упр рлось...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>это правила работы документами сотрудник захоч...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>почему бцж появляется шрамик</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>посмотрите одета это любому якутске происходило</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>ивашка простоквашкой</td>\n      <td>1.0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>это эрефия метода сделать говно запретить бону...</td>\n      <td>1.0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>честно говоря базовые потребности полностью по...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>жалко р ближайшие несколько лет кредиты брать ...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>8</th>\n      <td>выявлен каким анализам случайно рентгену ежего...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>9</th>\n      <td>говорит регионы развиваются работа достойная з...</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(data)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}