Multi label text classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import re\n",
    "import matplotlib\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from nltk.corpus import stopwords\n",
    "stop_words = set(stopwords.words('english'))\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.pipeline import Pipeline\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"train 2.csv\", encoding = \"ISO-8859-1\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Number of comments in each category"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>category</th>\n",
       "      <th>number_of_comments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>toxic</td>\n",
       "      <td>15294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>severe_toxic</td>\n",
       "      <td>1595</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>obscene</td>\n",
       "      <td>8449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>threat</td>\n",
       "      <td>478</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>insult</td>\n",
       "      <td>7877</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>identity_hate</td>\n",
       "      <td>1405</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        category  number_of_comments\n",
       "0          toxic               15294\n",
       "1   severe_toxic                1595\n",
       "2        obscene                8449\n",
       "3         threat                 478\n",
       "4         insult                7877\n",
       "5  identity_hate                1405"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_toxic = df.drop(['id', 'comment_text'], axis=1)\n",
    "counts = []\n",
    "categories = list(df_toxic.columns.values)\n",
    "for i in categories:\n",
    "    counts.append((i, df_toxic[i].sum()))\n",
    "df_stats = pd.DataFrame(counts, columns=['category', 'number_of_comments'])\n",
    "df_stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,0,'category')"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgcAAAGHCAYAAAAk+fF+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAIABJREFUeJzt3XmYHGW5///3h4QlJOyEEUggLGEJ\nBBQim6gTQAhrOAc4wEFZDpojsiqKuCAo8BMVRFGM8hUkgBIQEVBAiOCAiKwihPUQ2RLWQFiSsBm4\nf3/UM1A19Mz0pHumujuf13X1Nd1PVT19d3XP9D3PVooIzMzMzDotVnYAZmZm1licHJiZmVmBkwMz\nMzMrcHJgZmZmBU4OzMzMrMDJgZmZmRU4OTCrkaTzJZ1S0nNL0q8kvSzpjjJiMLPW4+TAWo6kJyQ9\nL2loruyzkjpKDKu/bAt8ChgREVuUHUx/kzRKUkgaXHYs9SKpXdKssuMwy3NyYK1qMHB02UH0laRB\nfTxkTeCJiJjfH/FYZa2SnLTK67D6c3JgreoHwJclLd91Q6X/PiV1SPpsun+wpL9JOlPSK5Iek7RN\nKp8p6QVJB3WpdmVJ0yTNlXSTpDVzdW+Qts2R9Iik/8ptO1/SZEnXSJoPjK8Q72qSrkrHz5D0uVR+\nKPBLYGtJ8yR9u9KJkPQ5SQ+l2B6UtFkq3zC97lckPSBpjy5x/UzStanuv0n6kKQfpS6MhyV9JLf/\nE5K+Iuk+SfMlnSupLR0/V9KfJa2Q238rSbem575XUnuX9+Lk9JxzJV0vaeW0+eb085UU19aS1k3n\n/FVJL0q6pJvz0Pm+T5L0jKRnJR2b276YpOMl/UvSS5IulbRil2MPlfQUcGM3zzFR0j8lvZbqmZDK\nD8m9B49J+t9UPhS4FlgtvZ556f3uNpZ03IGSnkzbTkjnf4e0bcn0Pj2Tbj+StGTa1i5plqSvSnoO\n+JWk+yXtnqt78XQeP1zpNdoiIiJ8862lbsATwA7A5cApqeyzQEe6PwoIYHDumA7gs+n+wcAC4BBg\nEHAK8BRwNrAksCMwFxiW9j8/Pf5E2v5j4Ja0bSgwM9U1GNgMeBHYKHfsq8DHyJL1pSq8npuAnwFL\nAR8GZgPb52K9pYdzsQ/wNPBRQMC6ZK0NiwMzgK8DSwDbpdewfi6uF4HN0/PeCDwOHJg7J3/pcs5v\nA9qA1YEXgH8AH0nn5EbgxLTv6sBLwC7pNX8qPR6eey/+BawHDEmPT+vhvbsY+Ebn+QO27eZcdB57\ncXpfxqZzuUPafkx6DSNSzL8ALu5y7AXp2CEV6t8ivZefSrGsDmyQtu0KrJPeg08CrwObpW3twKwu\ndfUUyxhgHlmX0hLA6cC/c6/jO+nYVYDhwK3AybnnWgB8L9U7BDgOuCT33BOB6WX/HvtW7q30AHzz\nrd433k8ONk5/rIfT9+Tg0dy2sWn/tlzZS8CH0/3zgam5bcOAd4CRwL7AX7vE9wve/6I8H7igh9cy\nMtW1TK7su8D5uVh7Sg6uA46uUP5x4DlgsVzZxcBJubj+X27bkcBDXc7JK13O+QG5x78DJnc5/op0\n/6vAhRXiPCj3Xnwzt+0LwJ96eO8uAM4hG3fR0+ei89gNcmXfB85N9x8iJV3p8apkX7qDc8eu3UP9\nvwDOrPIzekXn+0Ll5KCnWL5FShTStqWBt3k/OfgXsEtu+05kXU+dz/U2uSQUWI0sMVw2Pb4MOG4g\nf2d9a7ybuxWsZUXE/cAfgeMX4vDnc/ffSPV1LRuWezwz97zzgDlkf3TXBLZMzeevSHoFOAD4UKVj\nK1gNmBMRc3NlT5L9V1qNkWRfFpXqnRkR7/ZQb9fX29Pr78v+awL7dDkn25J9AXZ6Lnf/9QrPlXcc\n2X/kd6Tukf/pYV8onu8nyc5FZ1y/z8X0EFli1tbNsV11d66RtLOk21LX0CtkrSYrV9q3ilhWo/h5\ne50sWe20WnpdlV4jwOyIeDN3/DPA34C9lHXD7Qz8uofYbBHgwSjW6k4ka94+I1fWOXhvaeC1dD//\nZb0wRnbekTQMWBF4huyP+E0R8akeju3p0qjPACtKWiaXIKxB1lVQjZlkzdmV6h0pabFcgrAG8H9V\n1luLmWQtB59biGM/cK4i4jmgcxzGtsCfJd0cETO6qWMk8HC6vwbZueiM638i4m9dD5A0qrvnz6l4\nrlN//+/IumSujIh/S7qCLKHprs6eYnkWWD/3eAiwUm6XZ8iSiwfS4/xr7O75ppC1rg0G/h4R1X6+\nrEW55cBaWvqCuAQ4Klc2m+zL9dOSBqX/NCt9gfbFLpK2lbQEcDJwe0TMJGu5WE/SZ9JAr8UlfVTS\nhlXGP5Osz/i7kpaStAlwKNX/Z/dLsoGZmyuzrrLBkreTJUnHpZjagd2BqX150QvpImB3STul879U\nGig3oopjZwPvAmt3FkjaJ3fsy2Rffu/0UMcJkpaWtBHZWJDOAYw/B05N5wdJwyVN7MPrOhc4RNL2\naUDh6pI2IBsXsGSKfYGkncnGrXR6HlhJ0nK5sp5iuYzs/G2TPm/f5v1EA7LuoW+mY1Ym64a4qJfY\nryAbD3M0WTeNLeKcHNii4Dtkg8jyPgd8haw5diOyL+Ba/IaslWIO2SC+AwDSf/s7AvuR/ff2HO8P\nBqvW/mR93s8AvycbrzCtmgMj4rfAqSm+uWRfAitGxNvAHmRNyC+SDXg8MCIe7q6uekkJz0SywZCz\nyf5L/gpV/D1KTeinAn9LTe5bkQ22vF3SPOAqsr78x3uo5iaywZg3AKdHxPWp/Mfp+OslzSUb1Ldl\nH17XHWTJxplkY11uAtZMn4GjgEvJkpf/Ts/TedzDZF/oj6XXtFpPsUTEA2RjOKYCz5K9ry8Ab6Uq\nTwHuAu4DppO1nPW4SFdEvEHWurEW2UBeW8QpoqdWMjOz1pC6Bh4HFo+IBeVGUz+pG+sVYHQvSVFv\n9XwLWC8iPl234KxpueXAzKzJSNo9dY0MJZvKOJ1sxsjC1rciWXfVOfWJ0JqdkwMzs+Yzkayb6Rlg\nNLBfLGQzsLJFtWYC10bEzb3tb4uGAelWkHQesBvwQkRsnCs/EjiCbFGOqyPiuFT+NbIs9h3gqIi4\nLpVPIOuLGwT8MiJOS+VrkfW/rUjWv/aZ1KdqZmZmfTRQLQfnAxPyBZLGk2W/m0TERmRNY0gaQzZ4\na6N0zM/SiOZBZCvU7Uy2Qtj+aV/IBnidGRGjyQb8HNrvr8jMzKxFDcg6BxFxc26ecKfDyJZEfSvt\n80Iqn0i22txbwOOSZpAtSwowIyIeA5A0FZgo6SGypV//O+0zBTgJmNxbXCuvvHKMGtU1rHLNnz+f\noUO7Dqy3rnyequdzVR2fp+r5XFWnEc/T3Xff/WJEDO9tvzIXQVoP+LikU4E3gS9HxJ1kK7Tdlttv\nFu+v2jazS/mWZIt/vJIbfZzf/wMkTQImAbS1tXH66afX4aXUz7x58xg2rKfF4Ax8nvrC56o6Pk/V\n87mqTiOep/Hjxz/Z+17lJgeDgRWAznnKl0pam+JiHp2Cyl0g0cP+FUXEOaQRuePGjYv29va+Rd3P\nOjo6aLSYGpHPU/V8rqrj81Q9n6vqNPN5KjM5mAVcnkbY3iHpXbK1xmeRW4qW7KpknUt/Vip/EVhe\n0uDUepDf38zMzPqozKmMV5CNFUDSemRLjL5ItirYfuma5GuRTdO5A7gTGC1prbRk6H7AVSm5+Auw\nd6r3IODKAX0lZmZmLWRAWg4kXUx2qdCVJc0iW2b2POA8SfeTXUL0oPRF/4CkS4EHyaY4Hh4R76R6\njiC7tOsg4Ly0jChkl4CdKukU4B6yNc7NzMxsIQzUbIX9u9lUcZnOiDiVbP30ruXXANdUKH+M92c0\nmJmZWQ28QqKZmZkVODkwMzOzAicHZmZmVuDkwMzMzAqcHJiZmVmBkwMzMzMrKHOFxJYx6vir61bX\nsWMXcHCd6nvitF3rUo+ZmS1a3HJgZmZmBU4OzMzMrMDJgZmZmRU4OTAzM7MCJwdmZmZW4OTAzMzM\nCpwcmJmZWYGTAzMzMytwcmBmZmYFTg7MzMyswMmBmZmZFTg5MDMzswInB2ZmZlbg5MDMzMwKnByY\nmZlZgZMDMzMzK3ByYGZmZgVODszMzKxgQJIDSedJekHS/RW2fVlSSFo5PZaksyTNkHSfpM1y+x4k\n6dF0OyhXvrmk6emYsyRpIF6XmZlZKxqoloPzgQldCyWNBD4FPJUr3hkYnW6TgMlp3xWBE4EtgS2A\nEyWtkI6ZnPbtPO4Dz2VmZmbVGZDkICJuBuZU2HQmcBwQubKJwAWRuQ1YXtKqwE7AtIiYExEvA9OA\nCWnbshHx94gI4AJgz/58PWZmZq1scFlPLGkP4OmIuLdLL8DqwMzc41mprKfyWRXKu3veSWStDLS1\ntdHR0bHwLyI5duyCmuvo1DakfvXV47U1qnnz5rX066snn6vq+DxVz+eqOs18nkpJDiQtDXwD2LHS\n5gplsRDlFUXEOcA5AOPGjYv29vbewu3VwcdfXXMdnY4du4AzptfnbXnigPa61NOIOjo6qMd7tyjw\nuaqOz1P1fK6q08znqazZCusAawH3SnoCGAH8Q9KHyP7zH5nbdwTwTC/lIyqUm5mZ2UIoJTmIiOkR\nsUpEjIqIUWRf8JtFxHPAVcCBadbCVsCrEfEscB2wo6QV0kDEHYHr0ra5krZKsxQOBK4s43WZmZm1\ngoGayngx8HdgfUmzJB3aw+7XAI8BM4D/B3wBICLmACcDd6bbd1IZwGHAL9Mx/wKu7Y/XYWZmtigY\nkDEHEbF/L9tH5e4HcHg3+50HnFeh/C5g49qiNDMzM/AKiWZmZtaFkwMzMzMrcHJgZmZmBU4OzMzM\nrMDJgZmZmRU4OTAzM7MCJwdmZmZW4OTAzMzMCpwcmJmZWYGTAzMzMytwcmBmZmYFTg7MzMyswMmB\nmZmZFTg5MDMzswInB2ZmZlbg5MDMzMwKnByYmZlZgZMDMzMzK3ByYGZmZgVODszMzKzAyYGZmZkV\nODkwMzOzAicHZmZmVuDkwMzMzAqcHJiZmVnBgCQHks6T9IKk+3NlP5D0sKT7JP1e0vK5bV+TNEPS\nI5J2ypVPSGUzJB2fK19L0u2SHpV0iaQlBuJ1mZmZtaKBajk4H5jQpWwasHFEbAL8H/A1AEljgP2A\njdIxP5M0SNIg4GxgZ2AMsH/aF+B7wJkRMRp4GTi0f1+OmZlZ6xqQ5CAibgbmdCm7PiIWpIe3ASPS\n/YnA1Ih4KyIeB2YAW6TbjIh4LCLeBqYCEyUJ2A64LB0/BdizX1+QmZlZCxtcdgDJ/wCXpPurkyUL\nnWalMoCZXcq3BFYCXsklGvn9P0DSJGASQFtbGx0dHbXGzrFjF/S+U5XahtSvvnq8tkY1b968ln59\n9eRzVR2fp+r5XFWnmc9T6cmBpG8AC4BfdxZV2C2o3MoRPexfUUScA5wDMG7cuGhvb+9LuBUdfPzV\nNdfR6dixCzhjen3elicOaK9LPY2oo6ODerx3iwKfq+r4PFXP56o6zXyeSk0OJB0E7AZsHxGdX+iz\ngJG53UYAz6T7lcpfBJaXNDi1HuT3NzMzsz4qbSqjpAnAV4E9IuL13KargP0kLSlpLWA0cAdwJzA6\nzUxYgmzQ4lUpqfgLsHc6/iDgyoF6HWZmZq1moKYyXgz8HVhf0ixJhwI/BZYBpkn6p6SfA0TEA8Cl\nwIPAn4DDI+Kd1CpwBHAd8BBwadoXsiTjS5JmkI1BOHcgXpeZmVkrGpBuhYjYv0Jxt1/gEXEqcGqF\n8muAayqUP0Y2m8HMzMxq5BUSzczMrMDJgZmZmRU4OTAzM7MCJwdmZmZW4OTAzMzMCpwcmJmZWYGT\nAzMzMytwcmBmZmYFTg7MzMyswMmBmZmZFTg5MDMzs4KqkgNJ+0vaMN1fX9LNkm6UtEH/hmdmZmYD\nrdqWg1OAOen+6WSXUL4Z+Fl/BGVmZmblqfaqjMMj4nlJSwHbAnsD/wZe7LfIzMzMrBTVJgezJa0L\njAXujIi3JC0NqP9CMzMzszJUmxycDNwNvAPsm8q2B+7tj6DMzMysPFUlBxFxvqRL0/3XU/HtwH79\nFZiZmZmVoy9TGYcAe0k6Lj0eTPUtD2ZmZtYkqp3K+EngEeAA4IRUPBqY3E9xmZmZWUmqbTn4EbBv\nREwAFqSy24Et+iUqMzMzK021ycGoiLgh3Y/0823crWBmZtZyqk0OHpS0U5eyHYDpdY7HzMzMSlbt\nf/7HAn+UdDUwRNIvgN2Bif0WmZmZmZWiqpaDiLgN2AR4ADgPeBzYIiLu7MfYzMzMrARVtRxIWhKY\nHRHfz5UtLmnJiHir36IzMzOzAVftmINpwOZdyjYHrqvmYEnnSXpB0v25shUlTZP0aPq5QiqXpLMk\nzZB0n6TNcscclPZ/VNJBufLNJU1Px5wlycs6m5mZLaRqk4OxZFMX8+4ANq3y+POBCV3KjgduiIjR\nwA3pMcDOZGsojAYmkdZSkLQicCKwJdkUyhM7E4q0z6TccV2fy8zMzKpUbXLwKtDWpawNmF/NwRFx\nM+9f8rnTRGBKuj8F2DNXfkFkbgOWl7QqsBMwLSLmRMTLZK0ZE9K2ZSPi7xERwAW5uszMzKyPqp2t\n8DvgN5KOAh4D1gF+CFxaw3O3RcSzABHxrKRVUvnqwMzcfrNSWU/lsyqUVyRpElkrA21tbXR0dNTw\nEjLHjl3Q+05VahtSv/rq8doa1bx581r69dWTz1V1fJ6q53NVnWY+T9UmB98AziDrSlgSeBP4FfD1\nfoip0niBWIjyiiLiHOAcgHHjxkV7e/tChFh08PFX11xHp2PHLuCM6fVZW+qJA9rrUk8j6ujooB7v\n3aLA56o6Pk/V87mqTjOfp2qnMr4ZEYcDQ4EPAcMi4oiIeLOG534+dQmQfr6QymcBI3P7jQCe6aV8\nRIVyMzMzWwhVX5VR0nLAR8kGJ46XtJ2k7Wp47quAzhkHBwFX5soPTLMWtgJeTd0P1wE7SlohDUTc\nEbgubZsraas0S+HAXF1mZmbWR9Wuc3AwcDYwD3g9tymAtas4/mKgHVhZ0iyyWQenAZdKOhR4Ctgn\n7X4NsAswIz3XIQARMUfSyUDnwkvfiYjOQY6Hkc2IGAJcm25mZma2EKrt3D4V2DsiFupLNyL272bT\n9hX2DeDwbuo5j2yFxq7ldwEbL0xsZmZmVlRtt8Jg4Pr+DMTMzMwaQ7XJwfeAb0qqeoyCmZmZNadq\nuxW+SDZL4ThJL+U3RMQadY/KzMzMSlNtcvDpfo3CzMzMGkZVyUFE3NTfgZiZmVljqGoMgaQlJZ0q\n6TFJr6ayHSUd0b/hmZmZ2UCrdoDhmWRTBQ/g/aWJHyBbX8DMzMxaSLVjDv4DWDci5kt6FyAinpbU\n7QWOzMzMrDlV23LwNl0SCUnDgZcq725mZmbNqtrk4LfAFElrwXsXSvopMLW/AjMzM7NyVJscfB14\nApgOLA88Snblw2/3T1hmZmZWll7HHKRVEbcFvhoRx6TuhBfTNRDMzMysxfTachAR7wJXRsRb6fFs\nJwZmZmatq9puhZslbdWvkZiZmVlDqHYq45PAtZKuBGby/loHRMS3+iMwMzMzK0e1ycEQ4Ip0f0Su\n3N0LZmZmLabaAYkXAn/rHHdgZmZmravPAxLNzMystXlAopmZmRV4QKKZmZkV1Dog0cz6wdgpY+tW\n12HDDuPIKUfWXM/0g6bXIRozawZVJQcRcUh/B2JmZmaNoarkQNLa3W2LiMfqF46ZmZmVrdpuhRlk\n4wyUK+scdzCorhGZmZlZqartVijMapD0IeBE4K/9EZSZmZmVp9qpjAUR8RxwDPDdWgOQ9EVJD0i6\nX9LFkpaStJak2yU9KukSSUukfZdMj2ek7aNy9XwtlT8iaada4zIzM1tULVRykKwPLF3Lk0taHTgK\nGBcRG5N1UewHfA84MyJGAy8Dh6ZDDgVejoh1gTPTfkgak47bCJgA/EySuzvMzMwWQlXJgaS/Sro5\nd7sLuB34YR1iGAwMkTSYLNl4FtgOuCxtnwLsme5PTI9J27eXpFQ+NSLeiojHycZIbFGH2MzMzBY5\niuj92kmSDupSNB+4NyIerTkA6WjgVOAN4HrgaOC21DqApJHAtRGxsaT7gQkRMStt+xewJXBSOuai\nVH5uOuayCs83CZgE0NbWtvnUqVNrfQlMf/rVmuvo1DYEnn+jPnWNXX25+lTUgObNm8ewYcPKDqPf\nPPjSg3Wra/ig4cx+Z3bN9YxZaUwdomlcrf6Zqiefq+o04nkaP3783RExrrf9qh2QOKX3vfpO0gpk\n//WvBbwC/BbYuVIInYd0s6278g8WRpwDnAMwbty4aG9v71vQFRx8/NU119Hp2LELOGN6tZNIevbE\nAe11qacRdXR0UI/3rlHVY9GiTocNO4zJ8ybXXM/0vVp7EaRW/0zVk89VdZr5PFW7zsHlZGMA/por\n+zhwdETsXcPz7wA8HhGzc8+zDbC8pMERsYBsRcZn0v6zgJHArNQNsRwwJ1feKX+Mmdki7+zP31i3\nulbZZn5d6jv859vVIRrrD9UOSPwkcGuXsr8D42t8/qeArSQtncYObA88CPwF6Ew6DgKuTPevSo9J\n22+MrF/kKmC/NJthLWA0cEeNsZmZmS2Sqm2/fhMYCryWKxsG/LuWJ4+I2yVdBvwDWADcQ9bkfzUw\nVdIpqezcdMi5wIWSZpC1GOyX6nlA0qVkicUC4PCIeKeW2MzMzBZV1SYH1wG/kPS/EfGapGWBnwJ/\nqjWAiDiRbEGlvMeoMNsgIt4E9ummnlPJBjaamZlZDartVjgWWBaYI+kFsv/alyNbCMnMzMxaSLWz\nFV4Gdk3LJo8EZqZVEs3MzKzFVDtbYUfgiYj4P+C5VLY+sEZETOvH+MzMzGyAVdutcDYwt0vZ3FRu\nZmZmLaTa5GCViHi2S9mzwIfqHI+ZmZmVrNrk4DFJXVeraAcer284ZmZmVrZqpzKeBFyerlnwL2Ad\n4JB0MzMzsxZSVctBRFwJ7Ei2ENKu6edOqdzMzMxaSNVX+ImIO/CSxGZmZi2v15YDSaMknS/paUlv\npZ9TJK09EAGamZnZwOoxOZC0Idl1D1YBvgHskX4OB+5K283MzKyF9NatcBpwdkSc0KX8/HRRpO8D\nu/dLZGZmZlaK3pKDT/D+JZK7OgNPZTQzM2s5vY05GET3l2X+d9puZmZmLaS35OBOul/L4GDgrrpG\nY2ZmZqXrrVvhBOC6dJGly8iWTF4V2Iesu2Gn/g3PzMzMBlqPLQcRcSvZ4kebAjcAD6efmwIT0nYz\nMzNrIb0ughQRfwc+IWkIsCLwckS83u+RmZmZWSn6skLiG8DT/RiLmZmZNYBqr8poZmZmiwgnB2Zm\nZlbQbXIg6Qe5+9sNTDhmZmZWtp5aDibl7l/R34GYmZlZY+hpQOK9ki4DHgSWlPSdSjtFxLf6JTIz\nMzMrRU8tB3sD/yRb9EjAyAq3EbUGIGl5SZdJeljSQ5K2lrSipGmSHk0/V0j7StJZkmZIuk/SZrl6\nDkr7Pyqpu+tBmJmZWS+6bTmIiBeAUwAkDY6I7pZRrtWPgT9FxN6SlgCWBr4O3BARp0k6Hjge+Cqw\nMzA63bYEJgNbSloROBEYBwRwt6SrIuLlforZzMysZVU1WyEiDpG0gqQDJX0t/Vyx1ieXtCzZlR/P\nTc/zdkS8AkwEpqTdpgB7pvsTgQsicxuwvKRVyZZxnhYRc1JCMA2YUGt8ZmZmiyJFRO87SVsDV5Mt\nn/wksAawIbBrWkFx4Z5c+jBwDtm4hk2Bu4GjgacjYvncfi9HxAqS/gicFhG3pPIbyFoU2oGlIqKz\npeME4I2IOL3Cc04iDbZsa2vbfOrUqQsb/numP/1qzXV0ahsCz79Rn7rGrr5cfSpqQPPmzWPYsGFl\nh9FvHnzpwbrVNXzQcGa/M7vmesasNKYO0TSuVv9MzX5qbt3qGjz0XRbMr30m/PA1lqlDNI2rET9T\n48ePvzsixvW2X7UrJP4I+EJEvPdNKmlf4CzgowsX4nvPvxlwZETcLunHZF0I3VGFsuih/IOFEeeQ\nJSSMGzcu2tvb+xRwJQcff3XNdXQ6duwCzphe9cKVPXrigPa61NOIOjo6qMd716iOnHJk3eo6bNhh\nTJ43ueZ6pu81vQ7RNK5W/0yd/fkb61bXKtvM54Vbh9Zczz4HttceTANr5s9UtanfesClXcouA9at\n8flnAbMi4vZcnZsBz6fuAtLPF3L7j8wdPwJ4podyMzMz66Nqk4NHgf26lO0D/KuWJ4+I54CZ6ZLQ\nANuTdTFcRXZJaNLPK9P9q4AD06yFrYBXI+JZ4DpgxzQuYgWyK0leV0tsZmZmi6pq26+PAf4o6Siy\nMQejyGYM7FaHGI4Efp1mKjwGHEKWtFwq6VDgKbJEBOAaYBdgBvB62peImCPpZODOtN93ImJOHWIz\nMzNb5FSVHETErZLWAXYFVgP+AFxTjy/giPgn2RTErravsG8Ah3dTz3nAebXGY2ZmtqjryyWbXwYu\n6sdYzMzMrAH4qoxmZmZW4OTAzMzMCpwcmJmZWUHVyYGkNfszEDMzM2sMfWk5uAcgTWc0MzOzFtXj\nbAVJd5Nd7+AeYFAqPols2WQzMzNrQb21HOwNXA+sCSwt6R/AkpLGS2rdq/qYmZktwnpLDhaLiMsi\n4nhgLtklk0W2quE/JT3a3wGamZnZwOptEaTfSFqD7HoHSwErAG9GxH8CSFqxn+MzMzOzAdZjchAR\nW0oaDIwFbgF+CiwjaTLwj3TzNQzMzMxaSK+zFSJiQUTcA7wdEZ8A5gMdZBde+l7/hmdmZmYDrepr\nKwBfTD8jIi4BLumHeMzMzKxkVa9zEBHnp7tr908oZmZm1gj6vHxyujqjmZmZtShfW8HMzMwKnByY\nmZlZgZMDMzMzK3ByYGZmZgVODszMzKzAyYGZmZkVODkwMzOzAicHZmZmVuDkwMzMzAqcHJiZmVlB\nQyQHkgZJukfSH9PjtSTdLulRSZdIWiKVL5kez0jbR+Xq+Foqf0TSTuW8EjMzs+bXEMkBcDTwUO7x\n94AzI2I08DJwaCo/FHg5ItYFzkz7IWkMsB+wETAB+JmkQQMUu5mZWUspPTmQNALYFfhleixgO+Cy\ntMsUYM90f2J6TNq+fdp/IjDKJ5WFAAAYZ0lEQVQ1It6KiMeBGcAWA/MKzMzMWosiotwApMuA7wLL\nAF8GDgZuS60DSBoJXBsRG0u6H5gQEbPStn8BWwInpWMuSuXnpmMu6/J0SJoETAJoa2vbfOrUqTW/\nhulPv1pzHZ3ahsDzb9SnrrGrL1efihrQvHnzGDZsWNlh9JsHX3qwbnUNHzSc2e/MrrmeMSuNqUM0\njavVP1Ozn5pbt7oGD32XBfNr/99y+BrL1CGaxtWIn6nx48ffHRHjettv8EAE0x1JuwEvRMTdkto7\niyvsGr1s6+mYYmHEOcA5AOPGjYv29vZKu/XJwcdfXXMdnY4du4AzptfnbXnigPa61NOIOjo6qMd7\n16iOnHJk3eo6bNhhTJ43ueZ6pu81vQ7RNK5W/0yd/fkb61bXKtvM54Vbh9Zczz4HttceTANr5s9U\nqckB8DFgD0m7AEsBywI/ApaXNDgiFgAjgGfS/rOAkcAsSYOB5YA5ufJO+WPMzMysD0odcxARX4uI\nEREximxA4Y0RcQDwF2DvtNtBwJXp/lXpMWn7jZH1i1wF7JdmM6wFjAbuGKCXYWZm1lLKbjnozleB\nqZJOAe4Bzk3l5wIXSppB1mKwH0BEPCDpUuBBYAFweES8M/Bhm5mZNb+GSQ4iogPoSPcfo8Jsg4h4\nE9inm+NPBU7tvwjNzMwWDaVPZTQzM7PG4uTAzMzMCpwcmJmZWYGTAzMzMytwcmBmZmYFTg7MzMys\nwMmBmZmZFTg5MDMzswInB2ZmZlbg5MDMzMwKnByYmZlZgZMDMzMzK3ByYGZmZgVODszMzKzAyYGZ\nmZkVODkwMzOzAicHZmZmVuDkwMzMzAqcHJiZmVmBkwMzMzMrcHJgZmZmBU4OzMzMrMDJgZmZmRU4\nOTAzM7MCJwdmZmZWUGpyIGmkpL9IekjSA5KOTuUrSpom6dH0c4VULklnSZoh6T5Jm+XqOijt/6ik\ng8p6TWZmZs2u7JaDBcCxEbEhsBVwuKQxwPHADRExGrghPQbYGRidbpOAyZAlE8CJwJbAFsCJnQmF\nmZmZ9U2pyUFEPBsR/0j35wIPAasDE4EpabcpwJ7p/kTggsjcBiwvaVVgJ2BaRMyJiJeBacCEAXwp\nZmZmLUMRUXYMAEgaBdwMbAw8FRHL57a9HBErSPojcFpE3JLKbwC+CrQDS0XEKan8BOCNiDi9wvNM\nImt1oK2tbfOpU6fWHPv0p1+tuY5ObUPg+TfqU9fY1ZerT0UNaN68eQwbNqzsMPrNgy89WLe6hg8a\nzux3Ztdcz5iVxtQhmsbV6p+p2U/NrVtdg4e+y4L5tf9vOXyNZeoQTeNqxM/U+PHj746Icb3tN3gg\ngumNpGHA74BjIuI1Sd3uWqEseij/YGHEOcA5AOPGjYv29vY+x9vVwcdfXXMdnY4du4AzptfnbXni\ngPa61NOIOjo6qMd716iOnHJk3eo6bNhhTJ43ueZ6pu81vQ7RNK5W/0yd/fkb61bXKtvM54Vbh9Zc\nzz4HttceTANr5s9U2WMOkLQ4WWLw64i4PBU/n7oLSD9fSOWzgJG5w0cAz/RQbmZmZn1U9mwFAecC\nD0XED3ObrgI6ZxwcBFyZKz8wzVrYCng1Ip4FrgN2lLRCGoi4YyozMzOzPiq7W+FjwGeA6ZL+mcq+\nDpwGXCrpUOApYJ+07RpgF2AG8DpwCEBEzJF0MnBn2u87ETFnYF6CmZlZayk1OUgDC7sbYLB9hf0D\nOLybus4DzqtfdGZmZoum0sccmJmZWWMpu1vBFjUn1Wl65frfhpMm1qeuk+o3FdXMrBW45cDMzMwK\nnByYmZlZgbsVzMzMkjP23a1udY3YaU/OmPyBhXr77NhL/liHaPrGLQdmZmZW4OTAzMzMCpwcmJmZ\nWYGTAzMzMytwcmBmZmYFTg7MzMyswMmBmZmZFTg5MDMzswInB2ZmZlbg5MDMzMwKnByYmZlZgZMD\nMzMzK3ByYGZmZgVODszMzKzAyYGZmZkVDC47ADOzWjy0wYZ1qefNI4/goc8fVpe6Nnz4obrUY1YW\ntxyYmZlZgZMDMzMzK3ByYGZmZgVODszMzKygpZIDSRMkPSJphqTjy47HzMysGbVMciBpEHA2sDMw\nBthf0phyozIzM2s+LZMcAFsAMyLisYh4G5gKTCw5JjMzs6ajiCg7hrqQtDcwISI+mx5/BtgyIo7o\nst8kYFJ6uD7wyIAG2ruVgRfLDqIJ+DxVz+eqOj5P1fO5qk4jnqc1I2J4bzu10iJIqlD2gcwnIs4B\nzun/cBaOpLsiYlzZcTQ6n6fq+VxVx+epej5X1Wnm89RK3QqzgJG5xyOAZ0qKxczMrGm1UnJwJzBa\n0lqSlgD2A64qOSYzM7Om0zLdChGxQNIRwHXAIOC8iHig5LAWRsN2eTQYn6fq+VxVx+epej5X1Wna\n89QyAxLNzMysPlqpW8HMzMzqwMmBmZmZFTg5MDMzswInB2a2SJL0sWrKzBZFTg5KJuk/JC2Xe7y8\npD3LjKlRSRoqabHc48UkLV1mTI1M0tCyY2hwP6myzABJ+1RTZiBpiKT1y46jFk4OyndiRLza+SAi\nXgFOLDGeRnYDkE8Glgb+XFIsDUvSNpIeBB5KjzeV9LOSw2oYkraWdCwwXNKXcreTyKZBW2Vfq7Js\nkSZpd+CfwJ/S4w9Laro1d1pmnYMmVilB8/tS2VIRMa/zQUTMc8tBRWcCO5EWAYuIeyV9otyQGsoS\nwDCy37NlcuWvAXuXElEDk7QzsAuwuqSzcpuWBRaUE1VDO4nsQoAdABHxT0mjygtn4fhLqHx3Sfoh\n2eWmAzgSuLvckBrWfEmbRcQ/ACRtDrxRckwNKSJmSoXLjbxTViyNJiJuAm6SdH5EPFl2PE3gGbK/\nSXtQ/Ns0F/hiKRE1tgUR8WqX37+m4+SgfEcCJwCXkF086nrg8FIjalzHAL+V1HnNjFWBfUuMp1HN\nlLQNEGkp8aNIXQxW8LqkHwAbAUt1FkbEduWF1Hgi4l7gXkkXRYRbCnp3v6T/BgZJGk32+3dryTH1\nmVdItKYiaXGyS20LeDgi/l1ySA1H0srAj4EdeD/hPDoiXio1sAYj6XqypPzLwOeBg4DZEfHVUgNr\nMJKmU+EKt50iYpMBDKfhpa7ObwA7pqLrgJMj4q3youo7JwclkfSjiDhG0h+ofGnpPUoIqyFJ2i4i\nbpT0n5W2R8TlAx2TNT9Jd0fE5pLu6/yCk3RTRHyy7NgaiaQ1e9rurpkiSftExG97K2t07lYoz4Xp\n5+mlRtEcPgncCOxeYVsATg5yJA0HPgeMIvc7HhH/U1ZMDaqz1elZSbuS9a2PKDGehuQv/z77GtA1\nEahU1tDcclAySatExAtdytaPiEfKismam6Rbgb+SDR57byBiRPyutKAakKTdyM7TSLL1DZYFvh0R\nTTftbCBImsv7rZxLAIsD8yNi2fKiahy5WR3/RdZd1WlZYExEbFFKYAvJLQfl+6ukEyLiUoA0//pQ\nYEy5YTUeSRcCR3SuC5GaO8+LiO3LjazhLO1+895FxB/T3VeB8WXG0gwiIj/tk7RYW1N94fWzZ4C7\naJFZHW45KJmkVcmu+f0m0EY2qvzY/Hx+y0j6X7Jfsi8BqwNfITtXfyg1sAYj6RTg1oi4puxYGpmk\n9YDJQFtEbCxpE2CPiDil5NCahqTbImKrsuNoJJIWb4WB0k4OGoCkw8n6pN4F9o+Iv5UcUsOStC3w\nF+BF4CMR8VzJITWc1Pw7FHg73QSEm3+LJN1ElmD+IiI+ksruj4iNy42sMXUZELwYMA74ZERsXVJI\nDSlNX/wuWetvfors2qUFtRDcrVAySdOAZ4GNyQZDnSfp5oj4crmRNR5JnyFbE+JAYBPgGkmHpHnY\nlnRt/rVuLR0Rd3RZrMbz+LuXHxC8AHgCmFhOKA3tV2RL4J9J1l11CFmC3lScHJTv7Ii4It1/JS1e\n4/XKK9sL2DYN4LxY0u+B84GPlBpVg1H2bXcAsFZEnCxpJLBqRNxRcmiN5kVJ65AG2UnamyxRtwoi\n4pCyY2gSQyLiBklKMz1OkvRXmuyaOe5WaACS2oCPpod3dJ29YN2TtEREvF12HI1E0mSyLqrtImJD\nSSsA10fER3s5dJEiaW2y8T7bAC8DjwMHeOpeZZK+D5xCtmT5n4BNgWMi4qJSA2swkv4GfBy4jGwK\n9tPAaRHRVFdp9FUZSybpv4A7gH3IpsDcnv6DsS4kjZD0e0mzJT0v6XfAKmXH1YC2jIjDyQa5EhEv\nk009syRd+ntcROwADAc2iIhtnRj0aMeIeA3YDZgFrEc2ZsOKjiG7YuxRwObAZ8hW32wq7lYo3zeA\nj3a2FqQFbP5MlnVa0a+A35AlUgCfTmWfKi2ixvRvSYN4v7l8OFlLgiUR8a6kI4BLI2J+2fE0icXT\nz12AiyNiTrNfXKg/RMSd6e48svEGTcnJQfkW69KN8BJu0enO8Ij4Ve7x+ZKOKS2axnUW8HtgFUmn\nkl2G+JvlhtSQpkn6MtmCNe8lCBExp7yQGtofJD1M1q3whZR0vllyTA0nTZH9CrAmxRVKm+qCXh5z\nULLUj7cpcHEq2he4z4vYfJCkP5MNQOw8V/sDh3gRpA+StAGwPdko6Rsiwldl7ELS4xWKo9mmnA2k\nNH7ltYh4J11gaFlPJy6SdC/wcz64Qund3R7UgJwclEzS94DbgW3J/pDfDGzl5OCDJK0B/BTYmqzJ\n/FbgqIh4qtTAGoykrYAHImJuerwM2fKtt5cbmTW7NJtqFMX/iC8oLaAG1HlBr7LjqJWTg5JJ+kdE\nbNal7L2rxNn7JH2s6wJRlcoWdZLuATaL9MudBt/d1fVzZv6y64u0fPk6wD95/z/iiIijyouqcUha\nMd09CniBrGvvvcs0N1t3lccclETSYcAXgLUl3ZfbtAzgL7vKfgJ0/YKrVLaoU+Sy/jT4zr/rXXT3\nZQc4OahsHFkLlP+jrOxuss9P5yjN/EyOAJqqu8p/MMrzG+BasmU2j8+Vz222DLO/SdqabC76cElf\nym1aFhhUTlQN7TFJR5FdNwCyJPSxEuNpVP6y65v7gQ/hhaIqioi1qtlP0qciYlp/x1MrJwclSVcW\nfJVsUJ31bAlgGNnnNb808GtkI/Gt6PNkMxa+SfYfyw3ApFIjakz+suublYEHJd1Bsbl8j/JCakrf\nAxo+OfCYA2saktbsaZEaST+JiCMHMiZrPpL+QJY0LQN8mGwRMn/Z9ULSJyuVR8RNAx1LM5N0T+eF\nvhqZWw6saVSxet3HBiSQBudlbnt1Olm/8PeAPXPlnWVWgZOAummK/8idHJi1nh0j4jhJ/0G2zO0+\nZJe5dnLA+19ykhbv+oUnaUg5UTUuSbdExLbpUuD5LzZfCryFOTkwaz1e5rYHninUNxGxbfrpS4HX\nxxNlB1ANJwfWSvwNmPEytz3zTCHrN5LuIl0HJl30rCAi/nPgo+o7D0i0piNpaKWL5Ug6OCLOLyGk\nhuNlbs3KIWldsgsu7Qt0JgrXN9uUWScH1jTSana/BIZFxBqSNgX+NyK+UHJoDUXSUmTN5tuS9RHf\nAkyOCLcemA2QtDLpbmTrjbwLnAf8uFlap3z1P2smZwI7kV25koi4F/hEqRE1pguAjchWj/wpsCFw\nYakRmS1CJG0CnAH8APgd2XosrwE3lhlXX3jMgTWViJjZZXDdO93tuwhbPyI2zT3+S7pSnJn1M0l3\nA68A5wLHR0TnGhq3S2qa6dZODqyZzExdCyFpCbILnPhSxB90j6StIuI2AElb4lH4ZgNln4goLFcu\naa2IeLxZBiOCxxxYE5G0MvBjYAeymQnXA0dHxEulBtYgJE0nG2OwOLA+8FR6vCbwYERsXGJ4ZouE\nbq6023SXcXbLgTUFSYOAz0TEAWXH0sB2y91fAfh4un8zWTOnmfUTSRuQjfVZTlK+hWBZYKlyolp4\nHpBoTSEi3gEmlh1HI4uIJ9MS03uSDUBcGRie7vt6AWb9a32yBH15YPfcbTPgcyXGtVDcrWBNQ9Kp\nwHLAJcB76xxExD9KC6oBpVX/tu5cC0LSUODvEbFJuZGZtT5JW0fE38uOo1buVrBmsk36+Z1cWQDb\nlRBLIxPFWRzv4NUjzfqVpOMi4vvAf0vav+v2iDiqhLAWmpMDaxoRMb7sGJrEr8imTf0+Pd6TbFqV\nmfWfzplTd5UaRZ24W8GahqQ24P8DVouInSWNIWs+9xdfF5I2I1shUcDNEXFPySGZLRIk7RMRv+2t\nrNE5ObCmIelasv+KvxERm0oaDNwTEWNLDs3MDOh2KuMHyhqduxWsmawcEZdK+hpARCyQ5BUSzax0\nknYmu0z66pLOym1aFlhQTlQLz8mBNZP5klYiG4SIpK2AV8sNycwMgGfIxhvsAdydK58LfLGUiGrg\nbgVrGpI2B84CNgbuJ5vDv3dE3FdqYGZmiaTFI+LfZcdRKycH1lTSOIP1yQbaPdIKv4Rm1jrSxZVO\nIlu2fDDZ36qIiLXLjKuvnBxY00hXFrwEuCQi/lV2PGZmXUl6mKwb4W5y64002zVgnBxY05C0JrBv\nur1LlihcGhFPlRqYmVki6faI2LLsOGrl5MCakqTRwAnAARExqOx4zMwAJJ0GDAIuB97qLG+2Zd49\nW8GaiqRRwH+RtR68AxxXZjxmZl10thqMy5U13TLvbjmwpiHpdmBx4Ldk4w4eKzkkM7OW5OTAmoak\nDSLi4bLjMDPrTqss875Y2QGY9cHLks5NyygjaYykQ8sOysws53zgOmC19Pj/gGNKi2YhOTmwZnI+\nLfBLZ2YtbeWIuJRsRhURsYDiJdSbgpMDayYt8UtnZi2tJZZ592wFayYt8UtnZi3tS8BVwDqS/kZa\n5r3ckPrOAxKtaUjaDPgJvraCmTWwVljm3S0H1kzWAXYGRgJ7kc0n9mfYzEon6T+72bSeJCLi8gEN\nqEb+w2rN5ISI+K2kFYAdgDOAyby/6IiZWVl2Tz9XAbYBbkyPxwMdZCsmNg0PSLRm0jn4cFfg5xFx\nJbBEifGYmQEQEYdExCFkY6LGRMReEbEXsFHJoS0UJwfWTJ6W9Auy5ZOvkbQk/gybWWMZFRHP5h4/\nD6xXVjALywMSrWlIWhqYAEyPiEclrQqMjYjrSw7NzAwAST8FRgMXk7Ui7AfMiIgjSw2sj5wcmJmZ\n1VEanPjx9PDmiPh9mfEsDCcHZmZmVuDZCmZmZjWSdEtEbCtpLmmhts5NQETEsiWFtlDccmBmZmYF\nHultZmZmBU4OzMzMrMDJgZmZmRU4OTCzmkl6QtIOZcdhZvXh5MDMmla6+p2Z1ZmTAzMrkDRS0uWS\nZkt6SdJPJa0j6cb0+EVJv5a0fNr/QmAN4A+S5kk6LpVvJelWSa9IuldSe+451pJ0s6S5kv4s6WxJ\nF+W27yHpgXRsh6QNc9uekPRVSfcB8yV9RdLvuryGn0j6Uf+eKbPW5eTAzN4jaRDwR+BJYBSwOjCV\nbK72d4HVgA3JLpt9EkBEfAZ4Ctg9IoZFxPclrQ5cDZwCrAh8GfidpOHpqX4D3AGslOr5TC6G9ciW\nnj0GGA5cQ5Z45C+ytT/ZBbiWBy4CJuSSlcHAvsCF9TkrZoseJwdmlrcFWQLwlYiYHxFvRsQtETEj\nIqZFxFsRMRv4IfDJHur5NHBNRFwTEe9GxDTgLmAXSWsAHwW+FRFvR8QtwFW5Y/cFrk7P92/gdGAI\n2WVwO50VETMj4o10kZubgX3StgnAixFxd81nw2wR5eTAzPJGAk9GxIJ8oaRVJE2V9LSk18j+W1+5\nh3rWBPZJ3QKvSHoF2BZYlSz5mBMRr+f2n5m7vxpZywUAEfFu2r56N/sDTCFLSEg/3WpgVgMnB2aW\nNxNYo8JAv++SLQm7SVoG9tNkXQ2dui61OhO4MCKWz92GRsRpwLPAiukqm51G5u4/Q5ZcACBJafvT\nPTzfFcAmkjYGdgN+XcVrNbNuODkws7w7yL68T5M0VNJSkj4GLAPMA15J4wm+0uW454G1c48vAnaX\ntJOkQamedkkjIuJJsi6GkyQtIWlrYPfcsZcCu0raXtLiwLHAW8Ct3QUdEW8Cl5HGMkTEUzWcA7NF\nnpMDM3tPRLxD9kW9Ltkgw1lkYwC+DWwGvEo20PDyLod+F/hm6kL4ckTMBCYCXwdmk7UkfIX3/+Yc\nAGwNvEQ2aPESsgSAiHiErGXiJ8CLKZ7dI+LtXsKfAozFXQpmNfOFl8ysdJIuAR6OiBNrqGMN4GHg\nQxHxWt2CM1sEueXAzAacpI+mtRMWkzSBrJXhihrqWwz4EjDViYFZ7by6mJmV4UNkXRMrkXVdHBYR\n9yxMRZKGko15eJJsGqOZ1cjdCmZmZlbgbgUzMzMrcHJgZmZmBU4OzMzMrMDJgZmZmRU4OTAzM7OC\n/x/aAItfzbTPkwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1123a0940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_stats.plot(x='category', y='number_of_comments', kind='bar', legend=False, grid=True, figsize=(8, 5))\n",
    "plt.title(\"Number of comments per category\")\n",
    "plt.ylabel('# of Occurrences', fontsize=12)\n",
    "plt.xlabel('category', fontsize=12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Multi-Label\n",
    "\n",
    "How many comments have multiple labels?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5,0,'# of categories')"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg4AAAFQCAYAAADX1/YjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZoG8yiNrAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a368e56a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rowsums = df.iloc[:,2:].sum(axis=1)\n",
    "x=rowsums.value_counts()\n",
    "\n",
    "#plot\n",
    "plt.figure(figsize=(8,5))\n",
    "ax = sns.barplot(x.index, x.values)\n",
    "plt.title(\"Multiple categories per comment\")\n",
    "plt.ylabel('# of Occurrences', fontsize=12)\n",
    "plt.xlabel('# of categories', fontsize=12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vast majority of the comment texts are not labeled."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The distribution of the number of words in comment texts."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a33792cc0>"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYcAAAD8CAYAAACcjGjIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a32d15668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lens = df.comment_text.str.len()\n",
    "lens.hist(bins = np.arange(0,5000,50))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Most of the comment text length are within 500 characters, with some outliers up to 5,000 characters long."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Percentage of comments that are not labelled:\n",
      "0.8983211235124177\n"
     ]
    }
   ],
   "source": [
    "print('Percentage of comments that are not labelled:')\n",
    "print(len(df[(df['toxic']==0) & (df['severe_toxic']==0) & (df['obscene']==0) & (df['threat']== 0) & (df['insult']==0) & (df['identity_hate']==0)]) / len(df))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is no missing comment in comment text column."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of missing comments in comment text:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print('Number of missing comments in comment text:')\n",
    "df['comment_text'].isnull().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Have a peek the first comment, the text needs clean."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"Explanation\\rWhy the edits made under my username Hardcore Metallica Fan were reverted? They weren't vandalisms, just closure on some GAs after I voted at New York Dolls FAC. And please don't remove the template from the talk page since I'm retired now.89.205.38.27\""
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['comment_text'][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "categories = ['toxic', 'severe_toxic', 'obscene', 'threat', 'insult', 'identity_hate']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create a function to clean the text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_text(text):\n",
    "    text = text.lower()\n",
    "    text = re.sub(r\"what's\", \"what is \", text)\n",
    "    text = re.sub(r\"\\'s\", \" \", text)\n",
    "    text = re.sub(r\"\\'ve\", \" have \", text)\n",
    "    text = re.sub(r\"can't\", \"can not \", text)\n",
    "    text = re.sub(r\"n't\", \" not \", text)\n",
    "    text = re.sub(r\"i'm\", \"i am \", text)\n",
    "    text = re.sub(r\"\\'re\", \" are \", text)\n",
    "    text = re.sub(r\"\\'d\", \" would \", text)\n",
    "    text = re.sub(r\"\\'ll\", \" will \", text)\n",
    "    text = re.sub(r\"\\'scuse\", \" excuse \", text)\n",
    "    text = re.sub('\\W', ' ', text)\n",
    "    text = re.sub('\\s+', ' ', text)\n",
    "    text = text.strip(' ')\n",
    "    return text"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Clean up comment_text column "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['comment_text'] = df['comment_text'].map(lambda com : clean_text(com))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Much better!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'explanation why the edits made under my username hardcore metallica fan were reverted they were not vandalisms just closure on some gas after i voted at new york dolls fac and please do not remove the template from the talk page since i am retired now 89 205 38 27'"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['comment_text'][0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split to train and test sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [],
   "source": [
    "train, test = train_test_split(df, random_state=42, test_size=0.33, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(106912,)\n",
      "(52659,)\n"
     ]
    }
   ],
   "source": [
    "X_train = train.comment_text\n",
    "X_test = test.comment_text\n",
    "print(X_train.shape)\n",
    "print(X_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pipeline\n",
    "\n",
    "scikit-learn provides a Pipeline utility to help automate machine learning workflows. Pipelines are very common in Machine Learning systems, since there is a lot of data to manipulate and many data transformations to apply. So we will utilize pipeline to train every classifier."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OneVsRest multilabel strategy\n",
    "\n",
    "The Multi-label algorithm accepts a binary mask over multiple labels. The result for each prediction will be an array of 0s and 1s marking which class labels apply to each row input sample."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Naive Bayes\n",
    "\n",
    "OneVsRest strategy can be used for multilabel learning, where a classifier is used to predict multiple labels for instance. Naive Bayes supports multi-class, but we are in a multi-label scenario, therefore, we wrapp Naive Bayes in the OneVsRestClassifier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define a pipeline combining a text feature extractor with multi lable classifier\n",
    "NB_pipeline = Pipeline([\n",
    "                ('tfidf', TfidfVectorizer(stop_words=stop_words)),\n",
    "                ('clf', OneVsRestClassifier(MultinomialNB(\n",
    "                    fit_prior=True, class_prior=None))),\n",
    "            ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "... Processing toxic\n",
      "Test accuracy is 0.9191401279933155\n",
      "... Processing severe_toxic\n",
      "Test accuracy is 0.9900112041626312\n",
      "... Processing obscene\n",
      "Test accuracy is 0.9514802787747584\n",
      "... Processing threat\n",
      "Test accuracy is 0.9971135038644866\n",
      "... Processing insult\n",
      "Test accuracy is 0.9517271501547694\n",
      "... Processing identity_hate\n",
      "Test accuracy is 0.9910556600011394\n"
     ]
    }
   ],
   "source": [
    "for category in categories:\n",
    "    print('... Processing {}'.format(category))\n",
    "    # train the model using X_dtm & y\n",
    "    NB_pipeline.fit(X_train, train[category])\n",
    "    # compute the testing accuracy\n",
    "    prediction = NB_pipeline.predict(X_test)\n",
    "    print('Test accuracy is {}'.format(accuracy_score(test[category], prediction)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### LinearSVC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "SVC_pipeline = Pipeline([\n",
    "                ('tfidf', TfidfVectorizer(stop_words=stop_words)),\n",
    "                ('clf', OneVsRestClassifier(LinearSVC(), n_jobs=1)),\n",
    "            ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "... Processing toxic\n",
      "Test accuracy is 0.9599498661197516\n",
      "... Processing severe_toxic\n",
      "Test accuracy is 0.9906948479842003\n",
      "... Processing obscene\n",
      "Test accuracy is 0.9789019920621356\n",
      "... Processing threat\n",
      "Test accuracy is 0.9974173455629617\n",
      "... Processing insult\n",
      "Test accuracy is 0.9712299891756395\n",
      "... Processing identity_hate\n",
      "Test accuracy is 0.9919861752027194\n"
     ]
    }
   ],
   "source": [
    "for category in categories:\n",
    "    print('... Processing {}'.format(category))\n",
    "    # train the model using X_dtm & y\n",
    "    SVC_pipeline.fit(X_train, train[category])\n",
    "    # compute the testing accuracy\n",
    "    prediction = SVC_pipeline.predict(X_test)\n",
    "    print('Test accuracy is {}'.format(accuracy_score(test[category], prediction)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "... Processing toxic\n",
      "Test accuracy is 0.9548415275641391\n",
      "... Processing severe_toxic\n",
      "Test accuracy is 0.9910556600011394\n",
      "... Processing obscene\n",
      "Test accuracy is 0.9761104464573956\n",
      "... Processing threat\n",
      "Test accuracy is 0.9973793653506523\n",
      "... Processing insult\n",
      "Test accuracy is 0.9687612753755294\n",
      "... Processing identity_hate\n",
      "Test accuracy is 0.991758293928863\n"
     ]
    }
   ],
   "source": [
    "LogReg_pipeline = Pipeline([\n",
    "                ('tfidf', TfidfVectorizer(stop_words=stop_words)),\n",
    "                ('clf', OneVsRestClassifier(LogisticRegression(solver='sag'), n_jobs=1)),\n",
    "            ])\n",
    "for category in categories:\n",
    "    print('... Processing {}'.format(category))\n",
    "    # train the model using X_dtm & y\n",
    "    LogReg_pipeline.fit(X_train, train[category])\n",
    "    # compute the testing accuracy\n",
    "    prediction = LogReg_pipeline.predict(X_test)\n",
    "    print('Test accuracy is {}'.format(accuracy_score(test[category], prediction)))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}