sentiment_analysis_ign.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Dependencies\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import re\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Read the data from the data folder\n",
    "df = pd.read_csv('data/ign.csv')\n",
    "\n",
    "# Selecting what matters\n",
    "df = df[['score_phrase', 'title']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f16cbb4e690>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAw8AAAF9CAYAAABCqVQ1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzt3XmYHVWZ+PFvpzNsgSBhkX2HNyqIgbC4joMii4iOK4I4\nCDKigoLigiDqKCKCGy5sKiCiDg4Oi6K4gIgrID/WIW+zSgCBYIIJSdjS/fvj1CWXS3e6esu96Xw/\nz5Mnfe+pqnvOrapT9Z6lbldfXx+SJEmSNJgJ7c6AJEmSpGWDwYMkSZKkWgweJEmSJNVi8CBJkiSp\nFoMHSZIkSbUYPEiSJEmqxeBBkiRJUi0GD5IkSZJqMXiQJEmSVIvBgyRJkqRaJrY7AxHRCzwB9DW9\nfUZmfjAidgOOBwKYCZyYmec2rXsEcCiwHnAjcGRmXlulrQScAuwJrARcCRyamQ+PfakkSZKk8adT\neh62zsyVm/59MCLWBy4EvgWsDRwOnBYR0wEi4vXAccABwDrARcAlEbFKtc0TgO2AXYCtgF7grKVZ\nKEmSJGk86ZTgoauf9/YDbs3MszPzicy8ghIgHFylHwJ8NzOvyczHM/NkYBGwd0RMBA4EPpuZ92Xm\nI8CxwF4Rse6Yl0aSJEkahzolePhCRPwtIuZExOkRMQnYAbiuZbnrgR2rv7fvJ/0GYCdgC2D15vTM\n7AEWVtuVJEmSNESdEDxcC1wBbA28DHgJcCowBXikZdnZwFrV32sCcwZIn1K9bk2f07S+JEmSpCFo\n+4TpzNyp6eUtEfFR4GLgKvofztQ8sXo46bX19fX1dXWNaBOSJElSpxn2DW7bg4d+3A10UyY4r9mS\ntibwUPX3rH7S16I8dWlW0/ILmtKnNK0/qNmz5zNhgsGDJEmSxo811pg07HXbGjxExIuAt2Xm0U1v\nPw94HLgU+I+WVXYE/lL9fS0wHTi32lY3MA04E7iTMkRpOuURr0TENsCK1Xq19Pb20dvbN/iCkiRJ\n0nKgq6+vfTfHEbEBcCtwDHAGsCnwP8CvKI9avR34EHAesCvwY2DnzLw5InYHfgTsAdwEHAUcBERm\nPh4RJwC7AW+gTJQ+G1iQmW+rm79Zs+YZOUiSJGlcWXvt1YY9tKatE6Yz8z7gtcC+lKFGVwE/Az6W\nmbOAvYHDKBOnvwTsn5k3V+teBhwNnA/8A3gVsFdmPl5t/jjgz5QnMN0J/BN499IpmSRJkjT+tLXn\nodPZ8yBJkqTxZpnteZAkSZK07DB4kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbUY\nPEiSJEmqZWK7M6Blx/z58+npmdHubIyarbeeyqRJk9qdDUmSpGWGwYNq6+mZwfXfO5Kt1l2t3VkZ\nsdsemAfv/ArTpu3Q7qxIkiQtMwweNCRbrbsaL9xkjXZnoy3seZEkScs7gweppp6eGVz6o8PZeINl\nv+flnvvmwb5ft+dFkiQNicGDNAQbb7AaW2+2eruzIUmS1BY+bUmSJElSLQYPkiRJkmoxeJAkSZJU\ni8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIkSVItBg+SJEmSajF4kCRJ\nklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbUYPEiSJEmqxeBBkiRJUi0GD5IkSZJqMXiQ\nJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYPkiRJkmox\neJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIkSVItBg+SJEmS\najF4kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbVMbHcGmkXEV4APZuaE6vVuwPFA\nADOBEzPz3KbljwAOBdYDbgSOzMxrq7SVgFOAPYGVgCuBQzPz4aVXIkmSJGn86Jieh4h4EfBOoK96\nvT5wIfAtYG3gcOC0iJhepb8eOA44AFgHuAi4JCJWqTZ5ArAdsAuwFdALnLW0yiNJkiSNNx0RPETE\nBOA04EtAV/X2fsCtmXl2Zj6RmVdQAoSDq/RDgO9m5jWZ+XhmngwsAvaOiInAgcBnM/O+zHwEOBbY\nKyLWXXolkyRJksaPjggegPcAjwLnNb23A3Bdy3LXAztWf2/fT/oNwE7AFsDqzemZ2QMsrLYrSZIk\naYjaPuchIp5LGX70Chb3OgCsCdzTsvhsYK2m9DkDpE+pXremz2laf1ATJnQxYULX4AsuJ7q7OyXW\nHB3d3ROYOLF+mZb38kuSJLU9eAC+DJyembdFxKYtaf3dufeNML22KVMm0dVl8NAwefLKzG13JkbR\n5Mkrs8Yak4a0/Hgy1PJLkiS1NXiIiFdRhhEd1E/yLJ7dS7Am8FBT+pot6WtRnro0q2n5BU3pU5rW\nH9Ts2fPteWgyd+7CdmdhVM2du5A5c+YPafnxZKjllyRJ48NIGg/b3fOwP7ARcG9EQDUHIyJmUSZP\nv71l+R2Bv1R/XwtMB86t1ukGpgFnAndShihNpzzilYjYBlixWq+W3t4+env7Bl9wObFoUW+7szCq\nFi3q5amn6pdpeS+/JElSu4OHD1GegtSwEfAnyiNWu4GjI+JgykTqXSm/2bBzteypwI8i4gfATcBR\nwGPAzzKzNyLOAI6JiGsoE6VPAC7IzFlIkiRJGrK2Bg/VI1QfabyOiBWAvsy8v3q9N+WH3r4J3AXs\nn5k3V+teFhFHA+dTfufhamCvzHy82txxwGqUJzBNBC4B3rs0yiVJkiSNR+3ueXiGzLyb0uPQeH0V\nZSjSQMufRvl9iP7SngQOq/5JkiRJGiGf0yhJkiSpFoMHSZIkSbUYPEiSJEmqxeBBkiRJUi0GD5Ik\nSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYP\nkiRJkmoxeJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIkSVIt\nBg+SJEmSajF4kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbUYPEiSJEmqxeBBkiRJ\nUi0GD5IkSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGS\nJElSLQYPkiRJkmoxeJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXg\nQZIkSVItBg+SJEmSajF4kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbUYPEiSJEmq\nZWK7MxAR2wFfBnYAHgN+C3wwMx+MiN2A44EAZgInZua5TeseARwKrAfcCByZmddWaSsBpwB7AisB\nVwKHZubDS6lokiRJ0rjS1p6HiFgRuAy4HFgbeCElEDg1ItYHLgS+VaUdDpwWEdOrdV8PHAccAKwD\nXARcEhGrVJs/AdgO2AXYCugFzlo6JZMkSZLGn3YPW1oZ+ARwQmY+mZkPARcA2wBvB27NzLMz84nM\nvIISIBxcrXsI8N3MvCYzH8/Mk4FFwN4RMRE4EPhsZt6XmY8AxwJ7RcS6S7WEkiRJ0jjR1uAhMx/J\nzO9mZi9ARGwJvBP4IWUY03Utq1wP7Fj9vX0/6TcAOwFbAKs3p2dmD7Cw2q4kSZKkIWr7nAeAiNgE\nuA3oBs4EPgP8gjLPodlsYK3q7zWBOQOkT6let6bPaVp/UBMmdDFhQlfdxce97u52d1SNru7uCUyc\nWL9My3v5JUmSOiJ4yMy/AStUPQ+nU3oe+ui/Z6Sv6e/+7uwHS69typRJdHUZPDRMnrwyc9udiVE0\nefLKrLHGpCEtP54MtfySJEkdETw0ZObtEXEM8EfgUkrvQrM1gYeqv2f1k74W5alLs5qWX9CUPqVp\n/UHNnj3fnocmc+cubHcWRtXcuQuZM2f+kJYfT4ZafkmSND6MpPGwrcFDRLyG8jSlyMxF1duNnoOr\ngTe1rLIj8Jfq72uB6cC51ba6gWmUYU93UoYoTaca+hQR2wArVuvV0tvbR29v3+ALLicWLeptdxZG\n1aJFvTz1VP0yLe/llyRJanfPw9XAasAXIuJTwCTg08DvgFOBD0fEwcB5wK6U32zYuVr3VOBHEfED\n4CbgKMrvRPwsM3sj4gzgmIi4hjJR+gTggsxs9EpIUi3z58+np2dGu7MxKrbeeiqTJjlcTZI0PG0N\nHjLzkeqH4L5GGWr0KPAb4KDMnBURe1N+6O2bwF3A/pl5c7XuZRFxNHA+5Xcergb2yszHq80fRwlM\nbqCU8xLgvUutcJLGjZ6eGXz4p59htY3XaHdWRmTePXP40t6fYto0HzonSRqedvc8kJk3Av82QNpV\nlKFIA617GnDaAGlPAodV/yRpRFbbeA1W32qddmdDkqS28jmNkiRJkmoxeJAkSZJUi8GDJEmSpFoM\nHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIkSVItQw4eImLLAd5fMSJeOvIsSZIkSepE\nw+l5uHGA91cCfj6CvEiSJEnqYBPrLhgRbwLeDKwQET/oZ5HNgCdHK2OSJEmSOkvt4AFI4H6gC1iv\n+r/ZbOCgUcqXJEmSpA5TO3jIzJuBD0fERpn51jHMkyRJkqQONJSeBwAagUNErAWs0k/6PaOQL0mS\nJEkdZsjBQ0TsCXwXeG4/yX1A90gzJUmSJKnzDDl4AL4KXAWcDzw6utmRJEmS1KmGEzxsCLwoMxeO\ndmYkSZIkda7h/M5DDzB5tDMiSZIkqbMNp+fhcOCUiDgeuA1Y1JyYmU+MRsYkSZIkdZbhBA8XAasB\nb+knzQnTkiRJ0jg1nODhw6OeC0mSJEkdbzi/83D2GORDkiRJUocbzu88nEUZntSvzDxoRDmSJEmS\n1JGGM2zpeTwzeJgIbA48Bfx5NDIlSZIkqfMMZ9jSLq3vRcRE4HjgjtHIlCRJkqTOM5zfeXiWzHwK\n+AxwzGhsT5IkSVLnGZXgobIqsNYobk+SJElSBxnOhOkTePaE6UnA7sD1o5EpSZIkSZ1nOBOm386z\ng4eFwC3A0SPOkSRJkqSONJwJ05uOQT4kSZIkdbjh9DwQESsDewBbUXohEvhFZj4xinmTJEmS1EGG\nM+fhhcBlwHOBOdXbawB3RcTLM/P+UcyfJEmSpA4xnKctfRH4LbBBZq6ZmWsCGwE3ACePYt4kSZIk\ndZDhBA8vAd6bmX9vvJGZ9wHvAV4xWhmTJEmS1FmGEzw8CTzez/tzgVVGlh1JkiRJnWo4wcMtwFH9\nvH8UcOvIsiNJkiSpUw3naUtHA5dHxAHAjUAXsC2wKbDP6GVNkiRJUicZcs9DZv4BeB5wEdBN+XXp\nnwM7ZOYvRjd7kiRJkjrFkIOHiNgc+BlwdWb+e2buAcwE/jsiNhvtDEqSJEnqDMOZ8/A1ymNZf9/0\n3veBq4GvjEamJEmSJHWe4QQPLwUObnlU6wPA+4GXj1bGJEmSJHWW4QQPACv2895kyhwISZIkSePQ\ncJ629AvgnIj4JHAnJQB5AXACZS6EJEmSpHFoOD0PRwJTgOuAR4DZwFVAL2XokiRJkqRxaMg9D5n5\nIPDSiNgO2ApYBPRk5i2jnTlJkiRJnWM4w5YAyMwbKE9dkiRJkrQcGO6EaUmSJEnLGYMHSZIkSbUY\nPEiSJEmqxeBBkiRJUi0GD5IkSZJqGfbTlkZLRGwCfBV4OeW3In4OfCAz/xkRuwHHAwHMBE7MzHOb\n1j0COBRYD7gRODIzr63SVgJOAfYEVgKuBA7NzIeXVtkkSZKk8aQTeh4uAuYAGwPbAlOBkyNifeBC\n4FvA2sDhwGkRMR0gIl4PHAccAKxTbeeSiFil2u4JwHbALpTfo+gFzlpKZZIkSZLGnbYGDxExGbgW\n+FhmLqh+gO4c4BXA24FbM/PszHwiM6+gBAgHV6sfAnw3M6/JzMcz82TKD9btHRETgQOBz2bmfZn5\nCHAssFdErLtUCylJkiSNE20NHjJzbma+OzNnNb29KXAfsANwXcsq1wM7Vn9v30/6DcBOwBbA6s3p\nmdkDLKy2K0mSJGmI2j7noVk1JOn9wOuAj1HmOTSbDaxV/b0mZbhTf+lTqtet6XOa1h/UhAldTJjQ\nVXfxca+7uxNGuY2e7u4JTJxYv0zLe/mXZ+Np37vfJUkj0THBQ0S8FLiYMoTp8oj4GNDfnXtf09/D\nSa9typRJdHUZPDRMnrwyc9udiVE0efLKrLHGpCEtP54MtfzLs/G0793vkqSR6IjgISJeB5wLHJaZ\n36/ensWzewnWBB5qSl+zJX0tylOXZjUtv6ApfUrT+oOaPXu+PQ9N5s5d2O4sjKq5cxcyZ878IS0/\nngy1/Muz8bTv3e+SpJE0IrU9eIiIl1AmSb8pM3/TlHQt8K6WxXcE/tKUPp0SdBAR3cA04EzgTsoQ\npelUQ58iYhtgxWq9Wnp7++jt7Rt8weXEokW97c7CqFq0qJennqpfpuW9/Muz8bTvh7rf58+fT0/P\njDHM0dK19dZTmTTJnhdJGq62Bg/VU5G+TRmq9JuW5B8An4mIg4HzgF0pv9mwc5V+KvCjiPgBcBNw\nFPAY8LPM7I2IM4BjIuIaykTpE4ALWiZnS5KWoKdnBh+5+CxW3Wj9dmdlxB6deT8n7fMupk3zuRmS\nNFzt7nl4MeV3HU6JiFOa3u+j/DDc3pQfevsmcBewf2beDJCZl0XE0cD5lN95uBrYKzMfr7ZxHLAa\n5QlME4FLgPeOeYkkaZxZdaP1ec6Wm7Y7G5KkDtDW4CEzr2LJj4udSRmKNND6pwGnDZD2JHBY9U+S\nJEnSCPm8PkmSJEm1GDxIkiRJqsXgQZIkSVItBg+SJEmSajF4kCRJklRLux/VuszxB5MkSZK0vDJ4\nGKKenhnk939ArLdhu7MyYvn3e+Ed+/mDSZIkSarF4GEYYr0Nmbbp5u3OhiRJkrRUOedBkiRJUi0G\nD5IkSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElS\nLQYPkiRJkmoxeJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIk\nSVItBg+SJEmSajF4kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMHSZIkSbUYPEiSJEmqxeBB\nkiRJUi0GD5IkSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtUxsdwYkSepE\n8+fPp6dnRruzMWq23noqkyZNanc2JC3jDB4kSepHT88MPnbRT1hto43bnZURmzfzHk58/RuZNm2H\ndmdF0jLO4EGSpAGsttHGPGfLrdqdDUnqGM55kCRJklSLwYMkSZKkWgweJEmSJNVi8CBJkiSpFoMH\nSZIkSbUYPEiSJEmqxeBBkiRJUi0GD5IkSZJqMXiQJEmSVEtH/MJ0ROwBnANcnplvb0nbDTgeCGAm\ncGJmntuUfgRwKLAecCNwZGZeW6WtBJwC7AmsBFwJHJqZD495oSRJkqRxpu09DxHxceBkYAbQ15K2\nPnAh8C1gbeBw4LSImF6lvx44DjgAWAe4CLgkIlapNnECsB2wC7AV0AucNcZFkiRJksaltgcPwGxg\nJ+BOoKslbT/g1sw8OzOfyMwrKAHCwVX6IcB3M/OazHw8M08GFgF7R8RE4EDgs5l5X2Y+AhwL7BUR\n6459sSRJkqTxpe3BQ2aekZkLeHbgALADcF3Le9cDO1Z/b99P+g2UYGQLYPXm9MzsARZW25UkSZI0\nBB0x52EJpgD3tLw3G1ir+ntNYM4A6VOq163pc5rWX6IJE7qYMOGZMU139wQW1Vl5GdHdPYGJE+vF\nkN3dbY81R9VQyt5YfjwZavmXZ+Np33vcW+dJ0kh0evAA/fdI9I0wvZYpUybR1fXM1SdPXvlZ0ciy\nbPLklVljjUm1l507xvlZmoZS9sby48lQy788G0/73uN+aHXeeOI5L2k0dHrwMItn9xKsCTzUlL5m\nS/palKcuzWpafkFT+pSm9Zdo9uz5z+p5mDt3YZ1Vlxlz5y5kzpz5tZcdT4ZS9sby48lQy788G0/7\n3uPeOk+SRtKQ0EnBQx8tT1sCrgXe1fLejsBfmtKnA+cCREQ3MA04kzIBe06VPrNK3wZYsVpvUL29\nffT2PjNLixb11irMsmLRol6eeqpemZbnsjeWH0+GWv7l2Xja9x731nmSNBJtDx4iYsPqz0nAChGx\nAdCVmfcC5wGfiYiDq793pfxmw87VOqcCP4qIHwA3AUcBjwE/y8zeiDgDOCYirqFMlD4BuCAzG70S\nkiRJkmrqhJlT91T/3gzsQ+kl+BtAdZO/N3AY8AjwJWD/zLy5Sr8MOBo4H/gH8Cpgr8x8vNr2ccCf\nKU9guhP4J/DupVIqSZIkaZxpe89DZi4xgMnMqyhDkQZKPw04bYC0JymBx2EjyaMkSZKkzuh5kCRJ\nkrQMMHiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYP\nkiRJkmoxeJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1TGx3BiR1vvnz59PTM6Pd\n2Rg1W289lUmTJrU7G5IkLXMMHiQNqqdnBmddcBjrbrhqu7MyYg/c+yjvetM3mDZth3ZnRZKkZY7B\ng6Ra1t1wVTbeYvV2Z0OSJLWRcx4kSZIk1WLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYP\nkiRJkmoxeJAkSZJUi8GDJEmSpFoMHiRJkiTVYvAgSZIkqRaDB0mSJEm1GDxIkiRJqsXgQZIkSVIt\nBg+SJEmSajF4kCRJklSLwYMkSZKkWgweJEmSJNUysd0ZkCRJnWf+/Pn09MxodzZGzdZbT2XSpEnt\nzoa0zDN4kCRJz9LTM4NjLv49kzfast1ZGbG5M2/n+H1g2rQd2p0VaZln8CBJkvo1eaMtmbLlNu3O\nhqQO4pwHSZIkSbUYPEiSJEmqxeBBkiRJUi0GD5IkSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLw\nIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYPkiRJkmoxeJAkSZJUi8GDJEmSpFomtjsDYyki\nNgO+CewEPAqcD3w8M3vbmjFJktSx5s+fT0/PjHZnY9RsvfVUJk2a1O5saJwY18EDcAFwNfA2YB3g\nZ8CDwJfamSlJktS5enpmcOHFN7HhRlPbnZURu3fmDN6wD0ybtkO7s7JMMHAc3LgNHiJiOrAt8G+Z\nOQ+YFxFfBo7E4EGSJC3BhhtNZcstp7U7G1rKenpmcOvZlxPrbt7urIxYPnAnHDj6geO4DR6AHYC7\nM/OfTe9dD0RETMrM+W3KlyRJkjpUrLs50zZ+fruz0bHGc/CwJjCn5b3Z1f9rAYMGDxMmdDFhQtcz\n3uvunsD//f3eUclgu+Xf7+X53ROYOLHevPnu7gnc9sC8Mc7V0nHbA/PYYQhlh1L+e+4bH+W/5755\nbDfEff/AvY+Oca6WjgfufZTulwx938+7p7U6WfbMu2cO3dOGXvZHZ94/hrlaeh6deT/d04d23M+b\nec8Y52rpmDfzHrp33GnI+37uzNvHMFdLz9yZt9M9ff0h7ft7Z46PoSv3zpzBztO3G9K+B7juur+O\nUY6Wru23H1qre3f3BG594M4xys3SlQ/cyTbdmw953w+mq6+vb1Q32Cki4hPAv2fmjk3vbQn0AJtl\n5t/aljlJkiRpGTSeH9U6i9L70GxNoK9KkyRJkjQE4zl4uBbYOCKaA4gdgVsyc0Gb8iRJkiQts8bt\nsCWAiPgTcDPwIWADyqNaT87MU9uaMUmSJGkZNJ57HgDeDKwPPABcAZxj4CBJkiQNz7jueZAkSZI0\nesZ7z4MkSZKkUWLwIEmSJKkWgwdJkiRJtRg8SJIkSarF4EGSJElSLQYPy7CIODYiftvufKgzRMRK\nEdEbEa9od15GQ0T8W0TcGxE311j209XvuiwTIuLAiPh7u/PRLhExtTpWN46ITSJiYURs2e58dbKI\nODQi7mp3PkZqqPs7It4eEbMi4qc1lj07In448lwO+jmvrI7fFcb6s+oYrP6LiN9GxAlj8Lmeu0tR\nRMyIiIPanQ+Aie3OwLIsIl4D/AI4IzMPXdqfn5mfAz63tD+3rojYCjgWeDWwBjAH+DPw+cz86xh9\n5q7AP8dq+0PMy1IvfycZhfPjA8AfMvNto5uzsRcR04BPAP8KrAo8CFwMfC4zZ7Uzb0MREXcDawHP\nzcz5LWkfAk4G3pWZ5wz3MzLzb8DKI8hmR6q+u/WBRUAf8E/gcuCozFymA8eWsgE8DtwIHJeZv13S\nusPY3x8FzsrMj9ZYtq/6t0RV/k/IzNNb3t8DuDQzx1vDaq3vpVU/+/lBym9mfTEzb10a525ETACO\nyMwvj3A7dzPGddlY339k5tSx2O5wjLcTZGl7N+VXq/eNiJXanZlOEhEvAq4F7gOmZeYqwEuq11dF\nxI79rDMax+OHgOmjsJ0RGU75x6GRnh/PAe4c3SyNvYjYDfg9Zf9Prfb9q4HnAldHxDrtzN8wzAPe\n1M/77wAeYhg3JUvDKNUnI9EHHJaZK1fHwPaUY+CM9mZrVDSXbWVgXeBC4GcRsdkof9bqwB01l+2q\n/g1mWDfT7RAR3W38+Kf3MzAJ2I3yo7vXRsS/LaU8bA98ZDgrRkRrA/lY12XDvv/ogPpqSOx5GKaI\nWBPYh3Jg/5Tya9bfr9J+C/yqStsduItywL4feCflAD4kMy+rln8NcAKwFTAXOK3qVSAifgm8vOmj\n/wW4JzM3j4hPA7tn5osj4pWUyvttwNeADSk3MPtm5iNVBfQ14D+AhymtOccDx48k0l6CbwCXZOYn\nGm9UrRQfiIjbgCciYlPKzeF/AicCRwDnRsRbKa22W1JaOr6QmWdW38dKwNeB11Iqs/8HvDczb42I\ni4G9gNd8Q0mvAAAY+klEQVRExJszc7cxKFddg5X/SYCIeAGlPNOAp4AfA0dm5uNV+huAzwBbALOA\nL2fm16u0ScC3gT0o+/TopVO0wQ10fkTETMr++mm13BXAypm5S/X6X4ELgJspx/3Lqu/gUEqL10qZ\n+US17NnAipn59qVZtiWpLgCnAd/KzBMb72fmHZQg6jrKuf67lvX2Br4HvCIzb46I/YFjgI0oF7UT\nMvPbEfFy4NfAupk5p1p3Jcr+f3Nm/mKUi9RH6T16Z5W/Rn6fR+lNu7XpvcOA9wGbUOq8T2TmxVXa\nOsA5wEuBvwEnNa23KaUemJqZPRGxBqWuejWwGnAl8P7q/CEitgdOBbYB7gU+mZnnD6c+WZoy8+8R\n8b/AB6tyrFWV4xXACpT6+j2ZeX+VvjMl0NiiSvvD0s5zXZn5GPDliDgU2CMifswAZetnf98NfBZ4\nY7X8Q8ChmfmrKm1j4OsR8e/AjyjnwnqNz66ut3/KzFGt/5aUr6bF9omILwJTKOfluzJzXrX+sK5j\nTZ/9LeAQyvH/7iXdJzTl+VjK8fUU8J3MPHaAsg14rg4kM3uB24CjI6IP+HbVUHI7i/flQcDHKb0V\ns4CvZubXqs+cDnyFct4+BvwPpUdhUXUtO5VyLVsJuIlyv7QC5bjvjoiFwB6ZeeUgdc1vq3X2otSL\njfuAodRlXwD2q95P4AOZ+ccq7XXA54HNKPvhLMoIg4touf+IiE0o+/nFQDdwCSUYm1fds/2Uci/2\nBeB1EfGuKp/zqnw+BZycmV+oPvtuqh6z6lrzKWB/YD3glur7bORz7eo7fQ3wKKX37pgqbQpwCrAr\npWf8cuB9jbqnjmUq0ukw7wRuysz/owQNrePQ/pNyoq9LOQB+BfwVWBv4JdXFszpp/odyok+mVCaf\nqE5KMvM1Ta07a1Mq3VObPqc5Up4E7AvsDASwLaXygTIE5K2Ug3ga8HbKATfqrS8R8VxKK/s3+kvP\nzK9n5g1Nb70S2CQzz42InYDvUk6o1SjBzpcj4sXVsh+l3IBsQ+mCvJtyA01m7kO5MTmsnYFDzfJf\nHxErUo6FP1GOk5cA/wb8V7WdFwLnA5+ktL4dAnwhIvasNnUM8ELgBcBOlMCxUwx0flxOKSfVeOGt\ngfUiotH1/XLgN5n5SsoN9kmZ+Tz6b03sxNbD7SkXla8NkH4K5Wbk6bo3Irah3Fi/tQocNqteH5WZ\nq1EukqdFxNaZeRUwE3hL0zZ3p1xsfjnahalcDLwkIjZoeu8dlHoLoCsi3kS5kB1AOW8/CZwfERtW\ny3yVclOwEeUG4eAlfN63KS3021IaQR6jnAdExCqUC+5/U3qmDgXOiYjm7vxXUr8+GWtdVb67qv26\nP3BelfZFSh22GaWcK7L4utBN+X4vo9yYfgp4D513vLfqplzvTmSAsg3gKEoZp1AaCb4KkJmbsrhO\n32OAdceyHug3X032BXYAng88j3LNZyTXsUof5Rq9e2a+e7D7hMoLKN/zhpTGyiMj4s2tBVrCubrR\nEL6Xr1H27ZpN292IEuwekpmrVnn4VNVABiXw+wPlu9yF0qDUuC4cQTnft6bcsP8KODUzr6b0YD9Y\n3QddWaOuofruDunnPqBOXfbOKl8vAyYDP2dx/fMvVTlOqsr4ckqQsWfr/UdEdFWfN5MSAAewAWV4\nVMPEqsxrZ+aV1XtvBG6gHBcHA8dHxLZVWvOxfmRVzj0o9wffAy5pupaeSRlutj7lnnC/iHh3lXYO\nZbjZ86s8PUoJgmozeBi+g1kcvX4P+NeqRaXh95l5bWY+Smk5eCIzv5eZTwGXUloPqMbebUBpraS6\nqb6VchPS6nTgtsxsroSbb6q6KWMR/5mZ91FaexoX1b2A8zLz5sx8hNIiMml4RR9Uo9v69prLn1d9\nT1Aq2Usy85eZ2ZeZv6ecuAdU6Z8HdsrM2Zn5JPATSuXdSeqWf09gFeDTmfl4Zt5GqZQbQcBBwK8y\n8+LMXJSZv6a0XDTS/51Swd5ftUJ/YVRLMTL9nR+b0RQ8ADtSWktuoFRuUAUPNT+jzvCEpW0LYGFm\n3jtA+gxKRb82PN1DczHwkWr/kpl3AWtl5qXV68so4+WnVds4l3IT2vBG4L+rlsGxMJvSYveOKs9d\nlIvWeU3LHAR8OzP/mpm9mfm/lPpnvyr99cCXmuqmU/r7oKpF7A3AsZn5j8z8J6VVb8eqft2d0hr5\nlcx8MjOvoJR/QdNmhlKfjKUuSov5QmAhZejNo5QWZSiBz2szc0F1Hfgpi+uy6ZTGnc9l5hOZ+Rfg\nf+msY/7pvETEqhFxFOWG51LgvQxctlZ9wMXV9bJRp289tlmvbbB8fTEz51QttqdTrrMwOtexX2bm\nnVD7PqEX+K/qWvJHSuDZyE/zcTPQuVq7BzczHwQeoTR2NaxOuadcUC1zXWZOycxbqvQXUnoJ+7L0\nIv6pKf/PodzoPlZd6z6dmY1Aq/WYH6yuAbgmM6/tJ+uD1WV9lMauLTPznszso4wGWL9qFFyZ0giy\nsCrjHZm5WaOubjGdcnP+kcxcmGWu22can11ZATg9q5EGlXsy8zuZ+VRmXgRcD+zdz/YPotSpt1fL\nfoMyr/J11XVlb8r8ykczcyblvuG6qhf4tZTemkey9JQdDewWQxhS67ClYYiIXSg3/z8EyMzbI+LP\nlJ15HOUAnNm0ymNAc3fQAkoLQcN+wBERsTHl5FuB0gXW/JkHUbpOXzRI9pqfxrGQxZOZ1qPceDZk\nla+x0IiMnz6+ojwB6LLqZRfl+3l19fqepnW3AF5VDVVpmEA54aG00H89Il5GaXXoovOO47rl/yZw\nZ3XxaLgD2Liq1DajqSu1Kb1x870Bz5wT0DMquR+hAc6PPwHvorSufbNqWX0F8EfKZMuXRcRVlBap\n97Ul46NnSY0yjbReyvFxAXB/Zn63ZbnDqy7sdSnHy4osrjPOBT5ZtbQ9QLlI7D5Kee9PHyUA/Byl\nRfllwIKq96yxzOaUi88RTet1AbdUF7KVqXesblKt13zcN8a7b0qpH2ZWF3UAMvPn8PTwJxhafTKW\nGuPFz6jytzplSMn/q3oV1we+EhE7UL6fbsowLCitx3Myc27T9jri/K40AqNGS/xC4Drg1Zl5X9Wb\nNlDZ+tN83VpAGaayQlZDFNukjwHy1fTe/zX9fSelTobRuY7d0/J6sPuE21uuJXdSejZabcEA52o/\nyy7JRBZPpCZLr+lZwB8j4veUFvuzcvEDIvag1FtbVuv+C4sbmE6lNBrcGxG/oAzBvqD5PB9C/vt4\n9ndHU9pgddmqlH2zOyWoaay3YmY+GBGfA34QER+m7M9zqgaf/vLZDfyjadtQjqE1m1635jVbXv+N\nUlf0t/1Tms5BKMfFhpR7hwk0Hb9VL07j+gxwfUu+nmLxMNlBddpN17Li3ZSD4ramL39lYMOI+BTl\nQG5tBey3VTAiXk0Z3rIvcGGW8X9XtyzzPMpYwddl5j8GydtArY9dlIMDgMzsi4ixCh56KCfb8yk3\nN2Tm76gCmYj4D+DTTcs3V3gLKOPFPzjAtn9YLbNdlvGze1FatTpJ3fIP1N3eeG+gfdlIX5FnnsOd\n0pM44PlB6W5+gNKK/nLgS5Tj8mhgO+CRLPMD6mjnRMKBJLBiRGzeaDVsEZRxuA9Tuvzvp7QUvS4z\nLwGIiIMpwyX2AX5XnatPV+iZeWdE/IFyM3Ed8PAArWyj6VLKGOdplF6P77ekLwQ+lplfaV0xIhoX\nvjrHal/L/816q3+DHedDqU+WmqoX5b+q839fytCVK4G3ZebsiHgfiyeGtp7b0DnnN7QERs2qsdg/\nZeCy9WckvWbDuY95gtLr22p1ntmoNli+mtO7mtYdjevY08dxnfsEnn3ONOen2QIGOFfrqgKAVSk3\nz0+PhMjMgyPiJEpP4/7AR6shXCtQhhoeCZyZmY9HxPlN691ZDT18NfA6ypCbd1ACiuHk/8klpC2p\nLuuiNOptA7ykytfzKXPwGnn9VER8p8rbW6syviozWx+VuxB4NMsws2dpuja25rW/876/+nABcHDV\n89K67UaPTn91xsLq//Wzmjc3HJ1UGS0TImJVygFzCOVmp/HvRZShCLsxtPGXOwKZmRdUFcKqNHWN\nVuPXzqdMmvndQBup4UFKq15ju1uyOKoeVdUB+SvgwwMssqSbvjso3+fTImLDWPwkgh0pj/5s9OR0\n2pCloZT/TmDzahxlw1RKb0Rflf68lnWnsrgl9n6a9iklWGmrQc6PdSjnxxWUR5juTOm6vobSxfuv\nDDxkqXERbL7gbzHK2R+xzLyeEjw+a99XvUnvo1xE+4BZmbkf8DHgzCgTaKHMX/ldZl5ZBQ4bU4aD\nNPseZczwm3j2jfyoq1qA/5uyb/8d+EHLIv2dtxtXf86iBIh1jtW7WRx4NzSGXt5BOSc2bT5novxm\nxkD1wGD1STtMoNS9GwOnZObs6v3mMtwPTI6I5huPtp/fNT2XJZdtJB6jqQ5o6qEdqhkD5OnFlEfO\n1tU812YLFveujPZ1bIn3CZXN45lPF9qC8nQ/eOY9yZLO1bo+TfmeGsMDG/N6pmTmjCwPi9ieMkzo\nDZTGoscy8xtV4DCBxcMwG3M//yUzL8vMwyjBxz5RHp7QakT5r1GX7Qh8v6nx5xn7JiLWqoY0nZKZ\nL6M8fn3fAfK5avNw9ohYraXXoT+t17VN6b/Xrr/vofFZd1MC26lNabtGeejAXVXadk1p/9LUyFOL\nPQ9D9zZKBfa9zFzUnBDlSRqNiYB1x6beTemx2JASgZ5EOeGbJxo+yPB/z6GRj0spXZ7fprR6fh6Y\nP+BaI/cB4A9RfrDn45n5tyjjmd9IKcvlA6z37SqfB1JO6udTWmSOoExquhvYJSIuodyI7gqldbOq\niBcCW0bE5JYu/6WtTvkvp9xUHRcRn6VUEh+gTGaCEjT+PiJeSxnytCtlrGJjEthlwCFVC84iSktm\nuw12fhxE6Wr/DCVIWlCl3UU5dz7ftErzYxfvopTxLVXX+JsplezdY1aS4XsP8POqZ++EzHw4Ijan\nzGdZlTLB7w1UrZaZ+a2qUj+dEgzcRRny8BzKTebJlItH84TA8ynzBrZk6QXQ51KGItyUma1d7adT\nJuudT5m4/XLgwojYLTOvjojLKU8au5IyIfL9/X1AlifDXQZ8NiIaY7A/B1xeDYd5mFJvHRPliSg7\nU1pkd+5vewxen4y15nkBK1GCxynVZx9Nmbx5C+VG5gXAGtVyf6HMc/loVTfsSDn32zmMp65ZlH00\nUNlG4jZgtShPHrqCsh+H80NtJwG/jjJUsjHe/a2UB50M5WEbH46I91N6Vg+hzEuBkV/HWt3Fku8T\noAwDOjoiGjfuu1FuwuGZ9yNLPFf7+ezmY3gDFveK7tqy3X2BEyJi98xMFk9+vo1yTKwcEdtRhukc\nRwk8GsO8fgLMjDIU6TFKY9LDmTknypyh1SNiPcqY/sHyX+dxvUuqy+4GdqoaKHZk8cMpNqzK//Mo\nQ5qurvK/Pov3e/P9x80R8Ufga1GGnS+izHeaTP9zGBo2bTpu9qFMJO9vdMXplO/755THgr8J+G5E\nTM3MeyPiIsq9xf6U687plEbouRHxI+DEiHgjJcD7LGVYWX/D3Pplz8PQHUSJShf1k3YWZWevwTMj\n/f6GpzRe/5hyEP8fZdLPBZQnNuwXZWzdIVTj8qL8kuPCiFhQRdqt2+3vMxrvfYtysN9Aae09m1LB\nj8kky8zsoVQACyk30QsprT1vBA6vWly7WvNcVTpvp9wIP0KpaE/KzMaF/jDKSfIwZfLZWyhDN26u\nbs7PpNyY/HYsylVXnfJnmQT3WkqL+0OU4+BcqhvoLJMkD6ZMhJ5NeTrLflmeuAOlxfo2Skv3XyjH\n35K6a5eGOufHlZSb3t83pf2B0svSHFQ+ffxmmaD3McqN5MOUFsJz+1u23bI8NeMlVHNWImIB5TGO\nM4AXZ3lgQWt+D6IEDAdQxv/eQQkYLgK+XL13bEQcUn3GPMpE61sHGG87FuX6M+U4Pa8lqS/LZO+j\nKDfycymPJzy06WbkYMr5fh/ltz++wsB114GUG4sZlHrxn1QX8CwTC19NmQg6mzKB9MBcPClzqPXJ\nWPt6o94G/l7le4+qfjiUEkD8ndLb1KjXejJzISXAfD2lnMdRhvh1vCwPBRmwbNQ7T/tdJsuPb32F\n8sSb+ykNoFe0rDfo9rNMYH5lla+7KePK/wN4Y2Yu6ZG4zdteRBl+dB1ljs4NlMefj8Z1rNX/sOT7\nhD7K9b2Pco79L6Xh4tet+a5xrrZqHMOPUSbvrkuZ7N0YKtmoo39IqZN/ExGNSfJfysxLqrrjG5S6\n/8bq+/oQ5Sb9XMpQ1/WrvP+D0iPQGLL0a0rwdCdl6PZg+R/0GFhSXUZ51OwLKIHKpygB5aWUIVrz\nKI0/P6TcP/2pKuc3q/Vb7z/2o9xn3025VndR6rfmz2v9/Esp58xDlPu292V5amGr71TpP6HUkR8B\n3pCLH9bxriqP91B6R36Ui38U8XDKA11uoXznUynX5tq6+vo64nqrpSCaJqBVUfV8yqPgrljympI6\nTdWqdWZmDukRe5KkzlP1qq+UHfTbRQNx2NJyomrR/GKUp/7cTWkVmkMZby5pGRFlHtRhlCeotY7X\nlSQtmzrpUcxLZPCw/Pg+ZVjIFZQxd7dQurgeXeJakjrNPEq3/5vzmc8HlyQtuzpm+O1gHLYkSZIk\nqRYnTEuSJEmqxeBBkiRJUi0GD5IkSZJqMXiQJEmSVIvBgyRJkqRaDB4kSZIk1WLwIElaJkTE2RFx\nRbvzIUnLM4MHSdKyxB8nkqQ2MniQJA1bRHS3Ow8DiYiJ7c6DJI03VqySNA5FxFuAY4DNgaeAq4EP\nZGZPROwOfB6YCjwMfA84LjP7ImIC8FHgncAmwBzgx8DHM/PxiHglcDnwDuAk4ELg/RGxCfAl4MXA\nc4CbgWMy89c187spcGe13YOBXYB5wCmZ+fmmRbsi4j+rsq0LXAsckJl3Nm3jP4GPA7cCr4uIAE4G\nXk5pNEvgE5n5q+qz1wBOAV4NTAb+DpyRmV+s0lcGjgf2AdYH7gW+kZmn1CmbJI0n9jxI0jgTEesD\nP6TcvE8GNgZuAb4TEdsAFwPfoNzk70m52f5YtfoxwIeAdwOrUW6Y30q5+W72ZmDbzHx/RKwA/AqY\nC7wAWKP6/J9FxAuGmP1jKMHL6tX/n4uIXZvStwW2BJ4HbAGsDXyhZRv/Aeyama+rXv+4+n+DKm+/\nAn5SBQ1QAql1gedn5iTgLcAHIuI1VfpplMBjd2AScGiVr/8cYtkkaZlnz4MkjT+rUxqH/gmQmY8C\nHwaIiK8At2bmWdWy/xcRbwJ6q9cfAL6amX+sXl8XEd+gBBeHN33GOZk5u/p7T2AzYHpmzq3e+2pE\nHAy8HTh2CHk/NzP/Wv39vYj4CCVQubx67ylKL0gvsCAiLqXc1Df7n8y8p+n1TkBXZi6svoNzKD0T\n2wK/owQUTwCPAmTm/wM2rJadAuwPvD4z76i2d3lEfI/SO3PGEMomScs8gwdJGmcy89aI+BLwm4i4\niXKDfGFm/g7YijK0p3n53wNExOrAmsANLZu8BZgcEes0vXdH099TgW7gwTJC6GldVDfhQ3BLy+s7\nWrZxdxU4NDwGrNTPOs1eChwXEdsCqzS931jvc8BPgAeqpzldAfywCo62ogRiF0RE82TtLsrwJkla\nrhg8SNI4lJkfiYgvAnsBr6EMIbqA0nI/0CTnlQfZbPPN8xNNfy8E5mfmasPNb5PWvE1gca9Iax4G\n8nTeImJL4GeUoUdvzMx/RMTmwO2NZTLzZmDriHgp5bt6N/DZarjUwmqxlzb1iEjScss5D5I0zkRE\nV0RMycxZmXlOZu4PvJ4yzOYhynyB5uVfFRH7Ag9Shjpt17LJbYHZmTlrgI9MYFK0dDtUN+lDzn7L\n6y2A5iFIQ31U63RgBeBzmfmP6r0XP+MDI54TERMz8w+Z+anMnAbcCBxICTKeqrbTvM6G1VwPSVqu\nGDxI0vizH3BTROwcEROqpwW9lBIcnARsGhEfjoiVI2IqcBawaWb2AadTJgvvHBHdEfFiylyH05fw\neb8EbgJOi4iNImJiRLwVuKVafygOiIhpEbFCRBxICXTOb0rvGuL2bqv+f2W1zT0oE8ABNomILsqT\nqE6qhm1RBUEbAZmZCyhl/2RE7FR9JzsAvweOHGJeJGmZZ/AgSeNMZp5HeZrS9ymPO70HeAmwZ2be\nBryK8kjUfwC/oAQPJ1arHwt8m/L41keA71ACjuZJz89o/a+Cjn2A2ZQW+3mUCdb7Zuafhpj9rwFf\npDwi9vPAextzMqrPbe15aH2vNW9/BT4DfIvS63II5VGwP6A8nvU9Vd6nAndHxKPAJcA5lKFOAB+h\nPLHpQmABcAFwWmaeiCQtZ7r6+vyxTklSezX9RsMemfnLNmdHkjQAex4kSZIk1eLTliRJYyoibqH8\n0vVA+miZkCxJ6kwOW5IkSZJUi8OWJEmSJNVi8CBJkiSpFoMHSZIkSbUYPEiSJEmqxeBBkiRJUi0G\nD5IkSZJqMXiQJEmSVIvBgyRJkqRa/j9QinxIYse2pAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f16cbb4e110>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the distribuiton of categories\n",
    "\n",
    "# Adjusting size\n",
    "dims = (9, 4)\n",
    "fig, ax = plt.subplots(figsize=dims)\n",
    "\n",
    "# Plotting\n",
    "sns.countplot(x='score_phrase', data=df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>score_phrase</th>\n",
       "      <th>title</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>18625</td>\n",
       "      <td>18625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unique</th>\n",
       "      <td>11</td>\n",
       "      <td>12589</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>top</th>\n",
       "      <td>Great</td>\n",
       "      <td>Open Season</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freq</th>\n",
       "      <td>4773</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       score_phrase        title\n",
       "count         18625        18625\n",
       "unique           11        12589\n",
       "top           Great  Open Season\n",
       "freq           4773           10"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Statistics\n",
    "df.describe(include=['object', 'category'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Vocabulary is {u'Unbearable': 10, u'Great': 5, u'Good': 4, u'Okay': 8, u'Disaster': 3, u'Amazing': 0, u'Bad': 2, u'Masterpiece': 6, u'Awful': 1, u'Mediocre': 7, u'Painful': 9}\n"
     ]
    }
   ],
   "source": [
    "# This function clusterizes different categories into 2 classes: 0 - negative, 1 - positive\n",
    "def binarization(category):\n",
    "    positive = ['Amazing', 'Great', 'Good', 'Okay', 'Masterpiece']\n",
    "    negative = ['Awful', 'Mediocre', 'Bad', 'Painful', 'Unbearable', 'Disaster']\n",
    "    \n",
    "    if category in positive:\n",
    "        category = 1\n",
    "    \n",
    "    if category in negative:\n",
    "        category = 0\n",
    "    \n",
    "    return category\n",
    "\n",
    "positive = ['Amazing', 'Great', 'Good', 'Okay', 'Masterpiece']\n",
    "negative = ['Awful', 'Mediocre', 'Bad', 'Painful', 'Unbearable', 'Disaster']\n",
    "categories = positive + negative\n",
    "categories_dict = CountVectorizer(stop_words=None, lowercase=False).fit(categories).vocabulary_\n",
    "print \"Vocabulary is\", categories_dict\n",
    "\n",
    "# Classes to numbers\n",
    "def category_to_number(category):\n",
    "    return categories_dict[category]\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Create a new column with the binarized categories\n",
    "df['binarized'] = df['score_phrase'].apply(binarization)\n",
    "\n",
    "# Create a new column with numbers representing categories\n",
    "df['numbers'] = df['score_phrase'].apply(category_to_number)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>score_phrase</th>\n",
       "      <th>title</th>\n",
       "      <th>binarized</th>\n",
       "      <th>numbers</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Amazing</td>\n",
       "      <td>LittleBigPlanet PS Vita</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Amazing</td>\n",
       "      <td>LittleBigPlanet PS Vita -- Marvel Super Hero E...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Great</td>\n",
       "      <td>Splice: Tree of Life</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Great</td>\n",
       "      <td>NHL 13</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Great</td>\n",
       "      <td>NHL 13</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  score_phrase                                              title  binarized  \\\n",
       "0      Amazing                            LittleBigPlanet PS Vita          1   \n",
       "1      Amazing  LittleBigPlanet PS Vita -- Marvel Super Hero E...          1   \n",
       "2        Great                               Splice: Tree of Life          1   \n",
       "3        Great                                             NHL 13          1   \n",
       "4        Great                                             NHL 13          1   \n",
       "\n",
       "   numbers  \n",
       "0        0  \n",
       "1        0  \n",
       "2        5  \n",
       "3        5  \n",
       "4        5  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check out data\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f16cc00d190>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAigAAAF9CAYAAAAuppNcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMUS6WSsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f170c792110>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Check the binarized distribution. Classes are unbalanced!\n",
    "sns.countplot(x='binarized', data=df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# This function cleans the data and replace numbers by a common token: NUM. This function is inspired in kim's work\n",
    "# at https://github.com/yoonkim/CNN_sentence/blob/master/process_data.py\n",
    "\n",
    "def clean_str(string):\n",
    "    \"\"\"\n",
    "    Tokenization/string cleaning.\n",
    "    \"\"\"\n",
    "    string = re.sub(r\"[^A-Za-z0-9(),:!?\\'\\`]\", \" \", string)     \n",
    "    string = re.sub(r\"\\'s\", \" \\'s\", string) \n",
    "    string = re.sub(r\"\\'ve\", \" \\'ve\", string) \n",
    "    string = re.sub(r\"n\\'t\", \" n\\'t\", string) \n",
    "    string = re.sub(r\"\\'re\", \" \\'re\", string) \n",
    "    string = re.sub(r\"\\'d\", \" \\'d\", string) \n",
    "    string = re.sub(r\"\\'ll\", \" \\'ll\", string) \n",
    "    string = re.sub(r\",\", \" , \", string) \n",
    "    string = re.sub(r\"!\", \" ! \", string) \n",
    "    string = re.sub(r\"\\(\", \" \\( \", string) \n",
    "    string = re.sub(r\"\\)\", \" \\) \", string) \n",
    "    string = re.sub(r\":\", \" : \", string)\n",
    "    string = re.sub(r\"\\?\", \" \\? \", string) \n",
    "    string = re.sub(r\"\\s{2,}\", \" \", string)\n",
    "    string = string.strip().lower()\n",
    "    string = re.sub(r\"[0-9]{2,}\", \"NUM\", string)\n",
    "    return string\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Apply the function to the title column\n",
    "\n",
    "df['cleaned'] = df['title'].apply(clean_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>score_phrase</th>\n",
       "      <th>title</th>\n",
       "      <th>binarized</th>\n",
       "      <th>numbers</th>\n",
       "      <th>cleaned</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Amazing</td>\n",
       "      <td>LittleBigPlanet PS Vita</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>littlebigplanet ps vita</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Amazing</td>\n",
       "      <td>LittleBigPlanet PS Vita -- Marvel Super Hero E...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>littlebigplanet ps vita marvel super hero edition</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Great</td>\n",
       "      <td>Splice: Tree of Life</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>splice : tree of life</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Great</td>\n",
       "      <td>NHL 13</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>nhl NUM</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Great</td>\n",
       "      <td>NHL 13</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>nhl NUM</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  score_phrase                                              title  binarized  \\\n",
       "0      Amazing                            LittleBigPlanet PS Vita          1   \n",
       "1      Amazing  LittleBigPlanet PS Vita -- Marvel Super Hero E...          1   \n",
       "2        Great                               Splice: Tree of Life          1   \n",
       "3        Great                                             NHL 13          1   \n",
       "4        Great                                             NHL 13          1   \n",
       "\n",
       "   numbers                                            cleaned  \n",
       "0        0                            littlebigplanet ps vita  \n",
       "1        0  littlebigplanet ps vita marvel super hero edition  \n",
       "2        5                              splice : tree of life  \n",
       "3        5                                            nhl NUM  \n",
       "4        5                                            nhl NUM  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Take a look at the last 5\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The vocabulary has 7572 words\n",
      "The largest sentence has 14 tokens\n"
     ]
    }
   ],
   "source": [
    "# Instantiate an CountVectorizer object removing english stop words. Set lowercase false cause I've have done this\n",
    "# and I wanna keep NUM token in uppercase.\n",
    "vectorizer = CountVectorizer(stop_words='english', lowercase=False)\n",
    "\n",
    "# Fit the data \n",
    "titles_vectorized = vectorizer.fit(df['cleaned'])\n",
    "\n",
    "# Get the vocabulary\n",
    "vocabulary = titles_vectorized.vocabulary_\n",
    "\n",
    "words = set(vocabulary.keys())\n",
    "\n",
    "# Size of vocabulary\n",
    "size_vocab = len(vocabulary)\n",
    "print \"The vocabulary has {} words\".format(size_vocab)\n",
    "\n",
    "# Get the data cleaned and transform words in numerical tokens.\n",
    "text = df['cleaned'].tolist()\n",
    "text = [string.split() for string in text]\n",
    "# tokens is a list of lists. Each list inside tokens represents a sentence\n",
    "tokens = []\n",
    "for sentence in text:\n",
    "    dummy = []\n",
    "    for word in sentence:\n",
    "        if word in words:\n",
    "            dummy.append(vocabulary[word] + 1) # Sum 1 cause I want to leave the token 0 for padding in LSTM.\n",
    "    tokens.append(dummy)\n",
    "    \n",
    "# Maximum length of a sentence in the data\n",
    "max_len = max([len(sentence) for sentence in tokens])\n",
    "print \"The largest sentence has {} tokens\".format(max_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "14"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Gather info\n",
    "\n",
    "# Sentences\n",
    "X = tokens\n",
    "\n",
    "# binary labels\n",
    "y_2 = df['binarized']\n",
    "\n",
    "# All 11 classes\n",
    "y_11 = df['numbers']\n",
    "\n",
    "# Vocabulary size\n",
    "size_vocab\n",
    "\n",
    "# Maximum length\n",
    "max_len"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Let's split the data in training, validation and test\n",
    "# First split data into train and dummy\n",
    "X_train, X_dummy, y_2_train, y_2_dummy = train_test_split(X, y_2, train_size=0.6, random_state=7)\n",
    "\n",
    "# Now split dummy into validation and test\n",
    "X_validation, X_test, y_2_validation, y_2_test = train_test_split(X_dummy, y_2_dummy, test_size=0.5,\\\n",
    "                                                                  random_state=7)\n",
    "\n",
    "# Same procedure using all 11 labels\n",
    "X_train, X_dummy, y_11_train, y_11_dummy = train_test_split(X, y_11, train_size=0.6, random_state=7)\n",
    "\n",
    "# Now split dummy into validation and test\n",
    "X_validation, X_test, y_11_validation, y_11_test = train_test_split(X_dummy, y_11_dummy, test_size=0.5,\\\n",
    "                                                                  random_state=7)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "# Deep Learning Time\n",
    "\n",
    "# First using LSTM and 2 labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Dependencies\n",
    "import tflearn\n",
    "from tflearn.data_utils import to_categorical, pad_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Sequence padding\n",
    "trainX = pad_sequences(X_train, maxlen=max_len, value=0.) # max_len was found some cells above\n",
    "validationX = pad_sequences(X_validation, maxlen=max_len, value=0.)\n",
    "testX = pad_sequences(X_test, maxlen=max_len, value=0.)\n",
    "\n",
    "# Converting labels to binary vectors\n",
    "trainY_2 = to_categorical(y_2_train, nb_classes=2)\n",
    "validationY_2 = to_categorical(y_2_validation, nb_classes=2)\n",
    "testY_2 = to_categorical(y_2_test, nb_classes=2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Network building\n",
    "net = tflearn.input_data([None, max_len])\n",
    "net = tflearn.embedding(net, input_dim=size_vocab + 1, output_dim=128) # input_dim is size_vocab + 1 cause now we\n",
    "                                                                        # have a 0 token. \n",
    "net = tflearn.lstm(net, 128, dropout=0.8)\n",
    "net = tflearn.fully_connected(net, 2, activation='softmax')\n",
    "net = tflearn.regression(net, optimizer='adam', learning_rate=0.0001,\n",
    "                         loss='categorical_crossentropy')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m0.21728\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 0.21728 - acc: 0.9151 | val_loss: 0.57073 - val_acc: 0.7793 -- iter: 11175/11175\n",
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m0.21728\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 0.21728 - acc: 0.9151 | val_loss: 0.57073 - val_acc: 0.7793 -- iter: 11175/11175\n",
      "--\n"
     ]
    }
   ],
   "source": [
    "# Training\n",
    "model = tflearn.DNN(net, tensorboard_verbose=0)\n",
    "model.fit(trainX, trainY_2, validation_set=(validationX, validationY_2), show_metric=True,\n",
    "          batch_size=32, n_epoch = 10)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 0, 1, 1, 1, 1, 1, 1, 1, 1]\n"
     ]
    }
   ],
   "source": [
    "# Get the predictions -- List of list of probabilities\n",
    "y_pred_2 = model.predict(testX)\n",
    "\n",
    "# Get the list of classes\n",
    "y_pred_2 = np.array(y_pred_2)\n",
    "y_pred_2 = np.argmax(y_pred_2, axis=1).tolist()\n",
    "\n",
    "print y_pred_2[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# The classes are unbalanced. So accuracy is not a good metric for the problem. Let's use the confusion matrix and \n",
    "# and a classification report\n",
    "from sklearn.metrics import confusion_matrix, classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.51      0.41      0.46       852\n",
      "          1       0.83      0.88      0.86      2873\n",
      "\n",
      "avg / total       0.76      0.77      0.77      3725\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Classfication report\n",
    "print classification_report(y_2_test, y_pred_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[2535  338]\n",
      " [ 501  351]]\n"
     ]
    }
   ],
   "source": [
    "# Confusion matrix\n",
    "cm = confusion_matrix(y_2_test, y_pred_2, labels=[1,0])\n",
    "\n",
    "print cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# This function helps us to plot a fancy confusion matrix. It takes the confusion matrix as inpunt!\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,title='Confusion matrix', cmap=plt.cm.Blues, rotation=False):\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    bal = classes\n",
    "    tick_marks = np.arange(len(bal))\n",
    "    if rotation:\n",
    "        plt.xticks(tick_marks, bal, rotation=60)\n",
    "    else:\n",
    "        plt.xticks(tick_marks, bal)\n",
    "    plt.yticks(tick_marks, bal)\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgAAAAGaCAYAAAB5bftCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkCSphv4/8RwAVMXOcPoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f222858bf50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(cm, ['positive', 'negative'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Let's use CNN and 2 classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Dependencies\n",
    "from tflearn.layers.conv import conv_1d, global_max_pool\n",
    "from tflearn.layers.merge_ops import merge\n",
    "from tflearn.layers.core import input_data, dropout, fully_connected\n",
    "from tflearn.layers.estimator import regression\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m0.19644\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 0.19644 - acc: 0.9250 | val_loss: 0.70534 - val_acc: 0.7729 -- iter: 11175/11175\n",
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m0.19644\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 0.19644 - acc: 0.9250 | val_loss: 0.70534 - val_acc: 0.7729 -- iter: 11175/11175\n",
      "--\n",
      "[1, 1, 1, 1, 1, 1, 1, 1, 1, 1]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.48      0.42      0.45       852\n",
      "          1       0.83      0.86      0.85      2873\n",
      "\n",
      "avg / total       0.75      0.76      0.76      3725\n",
      "\n",
      "[[2484  389]\n",
      " [ 495  357]]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/lib/python2.7/dist-packages/matplotlib/collections.py:571: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  if self._edgecolors == str('face'):\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgAAAAGaCAYAAAB5bftCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdda7f6ac90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Network building\n",
    "network = tflearn.input_data([None, max_len])\n",
    "network = tflearn.embedding(network, input_dim=size_vocab + 1, output_dim=128) # input_dim is size_vocab + 1 cause now we\n",
    "                                                                        # have a 0 token. \n",
    "branch1 = conv_1d(network, 128, 3, padding='valid', activation='relu', regularizer=\"L2\")\n",
    "branch2 = conv_1d(network, 128, 4, padding='valid', activation='relu', regularizer=\"L2\")\n",
    "branch3 = conv_1d(network, 128, 5, padding='valid', activation='relu', regularizer=\"L2\")\n",
    "network = merge([branch1, branch2, branch3], mode='concat', axis=1)\n",
    "network = tf.expand_dims(network, 2)\n",
    "network = global_max_pool(network)\n",
    "network = dropout(network, 0.5)\n",
    "network = fully_connected(network, 2, activation='softmax')\n",
    "network = regression(network, optimizer='adam', learning_rate=0.001,\n",
    "                     loss='categorical_crossentropy', name='target')\n",
    "\n",
    "# Training\n",
    "model = tflearn.DNN(network, tensorboard_verbose=0)\n",
    "model.fit(trainX, trainY_2, validation_set=(validationX, validationY_2), show_metric=True,\n",
    "          batch_size=32, n_epoch = 10)\n",
    "\n",
    "\n",
    "# Get the predictions -- List of list of probabilities\n",
    "y_pred_2cnn = model.predict(testX)\n",
    "\n",
    "# Get the list of classes\n",
    "y_pred_2cnn = np.array(y_pred_2cnn)\n",
    "y_pred_2cnn = np.argmax(y_pred_2cnn, axis=1).tolist()\n",
    "\n",
    "print y_pred_2cnn[:10]\n",
    "\n",
    "# Classfication report\n",
    "print classification_report(y_2_test, y_pred_2cnn)\n",
    "\n",
    "# Confusion matrix\n",
    "cm = confusion_matrix(y_2_test, y_pred_2cnn, labels=[1,0])\n",
    "\n",
    "print cm\n",
    "\n",
    "plot_confusion_matrix(cm, ['positive', 'negative'])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# LSTM and 11 classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m1.50534\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 1.50534 - acc: 0.3933 | val_loss: 1.90316 - val_acc: 0.2617 -- iter: 11175/11175\n",
      "Training Step: 3500  | total loss: \u001b[1m\u001b[32m1.50534\u001b[0m\u001b[0m\n",
      "| Adam | epoch: 010 | loss: 1.50534 - acc: 0.3933 | val_loss: 1.90316 - val_acc: 0.2617 -- iter: 11175/11175\n",
      "--\n"
     ]
    }
   ],
   "source": [
    "# Converting labels to binary vectors\n",
    "trainY_11 = to_categorical(y_11_train, nb_classes=11)\n",
    "validationY_11 = to_categorical(y_11_validation, nb_classes=11)\n",
    "testY_11 = to_categorical(y_11_test, nb_classes=11)\n",
    "\n",
    "# Network building\n",
    "net = tflearn.input_data([None, max_len])\n",
    "net = tflearn.embedding(net, input_dim=size_vocab + 1, output_dim=128) # input_dim is size_vocab + 1 cause now we\n",
    "                                                                        # have a 0 token. \n",
    "net = tflearn.lstm(net, 128, dropout=0.8)\n",
    "net = tflearn.fully_connected(net, 11, activation='softmax')\n",
    "net = tflearn.regression(net, optimizer='adam', learning_rate=0.0001,\n",
    "                         loss='categorical_crossentropy')\n",
    "\n",
    "\n",
    "# Training\n",
    "model = tflearn.DNN(net, tensorboard_verbose=0)\n",
    "model.fit(trainX, trainY_11, validation_set=(validationX, validationY_11), show_metric=True,\n",
    "          batch_size=32, n_epoch = 10)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.00      0.00      0.00       348\n",
      "          1       0.00      0.00      0.00       132\n",
      "          2       0.15      0.12      0.13       247\n",
      "          3       0.00      0.00      0.00         2\n",
      "          4       0.28      0.22      0.25       956\n",
      "          5       0.37      0.42      0.39       956\n",
      "          6       0.00      0.00      0.00        16\n",
      "          7       0.21      0.27      0.24       380\n",
      "          8       0.22      0.44      0.29       597\n",
      "          9       0.00      0.00      0.00        72\n",
      "         10       0.00      0.00      0.00        19\n",
      "\n",
      "avg / total       0.23      0.27      0.24      3725\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdMAAAGaCAYAAABUqakJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCIKgQkKYBkEQBEGF/D/+rAyVYJd2xAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f168544b610>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get the predictions -- List of list of probabilities\n",
    "y_pred_11 = model.predict(testX)\n",
    "\n",
    "# Get the list of classes\n",
    "y_pred_11 = np.array(y_pred_11)\n",
    "y_pred_11 = np.argmax(y_pred_11, axis=1).tolist()\n",
    "\n",
    "# Classfication report\n",
    "print classification_report(y_11_test, y_pred_11)\n",
    "\n",
    "# Confusion matrix\n",
    "cm = confusion_matrix(y_11_test, y_pred_11, labels=[10,9,8,7,6,5,4,3,2,1,0])\n",
    "\n",
    "plot_confusion_matrix(cm, ['Unbearable', 'Painful', 'Okay', 'Mediocre', 'Masterpiece', \\\n",
    "                           'Great', 'Good', 'Disaster', 'Bad', 'Awful', 'Amazing'], rotation=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}