Chapter 8 - Movie Review Full Example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Chapter 8 - Movie Review Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "d = pandas.read_csv(\"data/movie_reviews.tsv\", delimiter=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Holdout split\n",
    "split = 0.7\n",
    "d_train = d[:int(split*len(d))]\n",
    "d_test = d[int((1-split)*len(d)):]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "vectorizer = CountVectorizer()\n",
    "features = vectorizer.fit_transform(d_train.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>producer</th>\n",
       "      <th>producer9and</th>\n",
       "      <th>producers</th>\n",
       "      <th>produces</th>\n",
       "      <th>producing</th>\n",
       "      <th>product</th>\n",
       "      <th>production</th>\n",
       "      <th>productions</th>\n",
       "      <th>productive</th>\n",
       "      <th>productively</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   producer  producer9and  producers  produces  producing  product  \\\n",
       "0         0             0          0         0          0        0   \n",
       "1         0             0          0         0          0        0   \n",
       "2         0             0          0         0          0        0   \n",
       "3         0             0          0         0          0        1   \n",
       "4         0             0          0         0          0        0   \n",
       "5         0             0          0         0          0        0   \n",
       "6         0             0          0         0          0        0   \n",
       "\n",
       "   production  productions  productive  productively  \n",
       "0           0            0           0             0  \n",
       "1           0            0           0             0  \n",
       "2           0            0           0             0  \n",
       "3           1            0           0             0  \n",
       "4           1            0           0             0  \n",
       "5           0            0           0             0  \n",
       "6           0            0           0             0  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "i = 45000\n",
    "j = 10\n",
    "words = vectorizer.get_feature_names()[i:i+10]\n",
    "pandas.DataFrame(features[j:j+7,i:i+10].todense(), columns=words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.21303978814816443"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "float(features.getnnz())*100 / (features.shape[0]*features.shape[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultinomialNB(alpha=1.0, class_prior=None, fit_prior=True)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB\n",
    "model1 = MultinomialNB()\n",
    "model1.fit(features, d_train.sentiment)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pred1 = model1.predict_proba(vectorizer.transform(d_test.review))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score, roc_auc_score, classification_report, roc_curve\n",
    "def performance(y_true, pred, color=\"g\", ann=True):\n",
    "    acc = accuracy_score(y_true, pred[:,1] > 0.5)\n",
    "    auc = roc_auc_score(y_true, pred[:,1])\n",
    "    fpr, tpr, thr = roc_curve(y_true, pred[:,1])\n",
    "    plot(fpr, tpr, color, linewidth=\"3\")\n",
    "    xlabel(\"False positive rate\")\n",
    "    ylabel(\"True positive rate\")\n",
    "    if ann:\n",
    "        annotate(\"Acc: %0.2f\" % acc, (0.1,0.8), size=14)\n",
    "        annotate(\"AUC: %0.2f\" % auc, (0.1,0.7), size=14)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEMCAYAAADAqxFbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSAwAAFDH/wFHaKwNaP9jSQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10a1c36d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "performance(d_test.sentiment, pred1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## tf-idf features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "vectorizer = TfidfVectorizer()\n",
    "features = vectorizer.fit_transform(d_train.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.5, 1)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEMCAYAAADAqxFbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nl19xAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x108c79550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pred2 = model1.predict_proba(vectorizer.transform(d_test.review))\n",
    "performance(d_test.sentiment, pred1, ann=False)\n",
    "performance(d_test.sentiment, pred2, color=\"b\")\n",
    "xlim(0,0.5)\n",
    "ylim(0.5,1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Parameter optimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "param_ranges = {\n",
    "    \"max_features\": [10000, 30000, 50000, None],\n",
    "    \"min_df\": [1,2,3],\n",
    "    \"nb_alpha\": [0.01, 0.1, 1.0]\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def build_model(max_features=None, min_df=1, nb_alpha=1.0, return_preds=False):\n",
    "    vectorizer = TfidfVectorizer(max_features=max_features, min_df=min_df)\n",
    "    features = vectorizer.fit_transform(d_train.review)\n",
    "    model = MultinomialNB(alpha=nb_alpha)\n",
    "    model.fit(features, d_train.sentiment)\n",
    "    pred = model.predict_proba(vectorizer.transform(d_test.review))\n",
    "    res = {\n",
    "        \"max_features\": max_features,\n",
    "        \"min_df\": min_df,\n",
    "        \"nb_alpha\": nb_alpha,\n",
    "        \"auc\": roc_auc_score(d_test.sentiment, pred[:,1])\n",
    "    }\n",
    "    if return_preds:\n",
    "        res['preds'] = pred\n",
    "    return res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'max_features': 10000, 'nb_alpha': 0.01, 'auc': 0.9452288421633197, 'min_df': 1}\n",
      "{'max_features': 10000, 'nb_alpha': 0.01, 'auc': 0.94534598458416985, 'min_df': 2}\n",
      "{'max_features': 10000, 'nb_alpha': 0.01, 'auc': 0.94528327613097241, 'min_df': 3}\n",
      "{'max_features': 10000, 'nb_alpha': 0.1, 'auc': 0.94535170553865855, 'min_df': 1}\n",
      "{'max_features': 10000, 'nb_alpha': 0.1, 'auc': 0.94544454696447144, 'min_df': 2}\n",
      "{'max_features': 10000, 'nb_alpha': 0.1, 'auc': 0.9453788017032404, 'min_df': 3}\n",
      "{'max_features': 10000, 'nb_alpha': 1.0, 'auc': 0.9451069192188557, 'min_df': 1}\n",
      "{'max_features': 10000, 'nb_alpha': 1.0, 'auc': 0.94514844185428892, 'min_df': 2}\n",
      "{'max_features': 10000, 'nb_alpha': 1.0, 'auc': 0.94509984639498434, 'min_df': 3}\n",
      "{'max_features': 30000, 'nb_alpha': 0.01, 'auc': 0.96420321713304125, 'min_df': 1}\n",
      "{'max_features': 30000, 'nb_alpha': 0.01, 'auc': 0.96356040650699282, 'min_df': 2}\n",
      "{'max_features': 30000, 'nb_alpha': 0.01, 'auc': 0.96297866511088048, 'min_df': 3}\n",
      "{'max_features': 30000, 'nb_alpha': 0.1, 'auc': 0.96332309138802907, 'min_df': 1}\n",
      "{'max_features': 30000, 'nb_alpha': 0.1, 'auc': 0.96286801427537505, 'min_df': 2}\n",
      "{'max_features': 30000, 'nb_alpha': 0.1, 'auc': 0.96222698001875662, 'min_df': 3}\n",
      "{'max_features': 30000, 'nb_alpha': 1.0, 'auc': 0.95663438781365639, 'min_df': 1}\n",
      "{'max_features': 30000, 'nb_alpha': 1.0, 'auc': 0.9564270685211389, 'min_df': 2}\n",
      "{'max_features': 30000, 'nb_alpha': 1.0, 'auc': 0.9559847525329952, 'min_df': 3}\n",
      "{'max_features': 50000, 'nb_alpha': 0.01, 'auc': 0.97030439304911353, 'min_df': 1}\n",
      "{'max_features': 50000, 'nb_alpha': 0.01, 'auc': 0.96733456459245626, 'min_df': 2}\n",
      "{'max_features': 50000, 'nb_alpha': 0.01, 'auc': 0.96336889820855887, 'min_df': 3}\n",
      "{'max_features': 50000, 'nb_alpha': 0.1, 'auc': 0.96838833698005011, 'min_df': 1}\n",
      "{'max_features': 50000, 'nb_alpha': 0.1, 'auc': 0.96585363413380598, 'min_df': 2}\n",
      "{'max_features': 50000, 'nb_alpha': 0.1, 'auc': 0.96251569229733958, 'min_df': 3}\n",
      "{'max_features': 50000, 'nb_alpha': 1.0, 'auc': 0.95877558564647003, 'min_df': 1}\n",
      "{'max_features': 50000, 'nb_alpha': 1.0, 'auc': 0.95769994171026562, 'min_df': 2}\n",
      "{'max_features': 50000, 'nb_alpha': 1.0, 'auc': 0.95611163233186192, 'min_df': 3}\n",
      "{'max_features': None, 'nb_alpha': 0.01, 'auc': 0.97338569521829166, 'min_df': 1}\n",
      "{'max_features': None, 'nb_alpha': 0.01, 'auc': 0.96733456459245626, 'min_df': 2}\n",
      "{'max_features': None, 'nb_alpha': 0.01, 'auc': 0.96336889820855887, 'min_df': 3}\n",
      "{'max_features': None, 'nb_alpha': 0.1, 'auc': 0.97083593241650989, 'min_df': 1}\n",
      "{'max_features': None, 'nb_alpha': 0.1, 'auc': 0.96585363413380598, 'min_df': 2}\n",
      "{'max_features': None, 'nb_alpha': 0.1, 'auc': 0.96251569229733958, 'min_df': 3}\n",
      "{'max_features': None, 'nb_alpha': 1.0, 'auc': 0.95977324565853028, 'min_df': 1}\n",
      "{'max_features': None, 'nb_alpha': 1.0, 'auc': 0.95769994171026562, 'min_df': 2}\n",
      "{'max_features': None, 'nb_alpha': 1.0, 'auc': 0.95611163233186192, 'min_df': 3}\n"
     ]
    }
   ],
   "source": [
    "from itertools import product\n",
    "results = []\n",
    "for p in product(*param_ranges.values()): \n",
    "    res = build_model(**dict(zip(param_ranges.keys(), p)))\n",
    "    results.append( res )\n",
    "    print res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "opt = pandas.DataFrame(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x108e2d150>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEICAYAAACuxNj9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nVv0/tQxDD2oRguEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109864c50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mf_idx = [0,9,18,27]\n",
    "plot(opt.max_features[mf_idx], opt.auc[mf_idx], linewidth=2)\n",
    "title(\"AUC vs max_features\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x10982b650>"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEICAYAAAC9E5gJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x105b4d850>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mdf_idx = [27,28,29]\n",
    "plot(opt.min_df[mdf_idx], opt.auc[mdf_idx], linewidth=2)\n",
    "title(\"AUC vs min_df\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x10a0e80d0>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEICAYAAAC9E5gJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEEIIgP8Pohr7ag9mlPUAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x104418910>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nba_idx = [27,30,33]\n",
    "plot(opt.nb_alpha[nba_idx], opt.auc[nba_idx], linewidth=2)\n",
    "title(\"AUC vs alpha\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.5, 1)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEMCAYAAADAqxFbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1076ce810>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pred3 = build_model(nb_alpha=0.01, return_preds=True)['preds']\n",
    "performance(d_test.sentiment, pred1, ann=False)\n",
    "performance(d_test.sentiment, pred2, color=\"b\", ann=False)\n",
    "performance(d_test.sentiment, pred3, color=\"r\")\n",
    "xlim(0,0.5)\n",
    "ylim(0.5,1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "vectorizer = TfidfVectorizer(strip_accents='unicode', stop_words='english', min_df=3, max_features=30000, norm=\"l2\")\n",
    "features = vectorizer.fit_transform(d_train.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEMCAYAAADAqxFbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10b7bc490>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model3 = MultinomialNB()\n",
    "model3.fit(features, d_train.sentiment)\n",
    "pred3 = model3.predict_proba(vectorizer.transform(d_test.review))\n",
    "performance(d_test.sentiment, pred3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=1,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "model2 = RandomForestClassifier(n_estimators=100)\n",
    "model2.fit(features, d_train.sentiment)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEMCAYAAADAqxFbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAJr5fyVJrBW2I6ooAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1075e3fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pred2 = model2.predict_proba(vectorizer.transform(d_test.review))\n",
    "performance(d_test.sentiment, pred2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Word2Vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import re, string\n",
    "stop_words = set(['all', \"she'll\", \"don't\", 'being', 'over', 'through', 'yourselves', 'its', 'before', \"he's\", \"when's\", \"we've\", 'had', 'should', \"he'd\", 'to', 'only', \"there's\", 'those', 'under', 'ours', 'has', \"haven't\", 'do', 'them', 'his', \"they'll\", 'very', \"who's\", \"they'd\", 'cannot', \"you've\", 'they', 'not', 'during', 'yourself', 'him', 'nor', \"we'll\", 'did', \"they've\", 'this', 'she', 'each', \"won't\", 'where', \"mustn't\", \"isn't\", \"i'll\", \"why's\", 'because', \"you'd\", 'doing', 'some', 'up', 'are', 'further', 'ourselves', 'out', 'what', 'for', 'while', \"wasn't\", 'does', \"shouldn't\", 'above', 'between', 'be', 'we', 'who', \"you're\", 'were', 'here', 'hers', \"aren't\", 'by', 'both', 'about', 'would', 'of', 'could', 'against', \"i'd\", \"weren't\", \"i'm\", 'or', \"can't\", 'own', 'into', 'whom', 'down', \"hadn't\", \"couldn't\", 'your', \"doesn't\", 'from', \"how's\", 'her', 'their', \"it's\", 'there', 'been', 'why', 'few', 'too', 'themselves', 'was', 'until', 'more', 'himself', \"where's\", \"i've\", 'with', \"didn't\", \"what's\", 'but', 'herself', 'than', \"here's\", 'he', 'me', \"they're\", 'myself', 'these', \"hasn't\", 'below', 'ought', 'theirs', 'my', \"wouldn't\", \"we'd\", 'and', 'then', 'is', 'am', 'it', 'an', 'as', 'itself', 'at', 'have', 'in', 'any', 'if', 'again', 'no', 'that', 'when', 'same', 'how', 'other', 'which', 'you', \"shan't\", 'our', 'after', \"let's\", 'most', 'such', 'on', \"he'll\", 'a', 'off', 'i', \"she'd\", 'yours', \"you'll\", 'so', \"we're\", \"she's\", 'the', \"that's\", 'having', 'once'])\n",
    "\n",
    "def tokenize(docs):\n",
    "    pattern = re.compile('[\\W_]+', re.UNICODE)\n",
    "    sentences = []\n",
    "    for d in docs:\n",
    "        sentence = d.lower().split(\" \") \n",
    "        sentence = [pattern.sub('', w) for w in sentence]\n",
    "        sentences.append( [w for w in sentence if w not in stop_words] )\n",
    "    return sentences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['all', \"she'll\", \"don't\", 'being', 'over', 'through', 'yourselves', 'its', 'before', \"he's\", \"when's\", \"we've\", 'had', 'should', \"he'd\", 'to', 'only', \"there's\", \"here's\", 'under', 'ours', 'has', \"haven't\", 'do', 'them', 'his', 'above', 'very', \"who's\", \"they'd\", 'cannot', \"you've\", 'they', 'not', 'during', 'him', 'nor', \"we'll\", 'did', \"they've\", 'this', 'she', 'each', \"won't\", 'where', \"mustn't\", \"isn't\", \"i'll\", \"why's\", 'because', \"you'd\", 'doing', 'theirs', 'some', \"hasn't\", 'are', 'further', 'ourselves', 'out', 'what', 'for', 'herself', 'below', 'does', \"shouldn't\", \"they'll\", 'between', 'be', 'we', 'after', \"doesn't\", 'here', 'hers', \"aren't\", 'by', 'both', 'about', 'her', 'of', 'could', 'against', \"i'd\", \"weren't\", \"i'm\", 'or', \"can't\", 'own', 'into', 'yourself', 'down', \"hadn't\", \"couldn't\", 'your', \"you're\", 'from', \"how's\", 'would', 'their', \"it's\", 'there', 'been', \"he'll\", 'whom', 'too', 'themselves', 'was', 'until', 'more', 'himself', \"i've\", 'am', \"what's\", 'but', 'it', 'with', 'than', 'those', 'he', 'me', \"they're\", 'myself', \"wasn't\", 'up', 'while', 'ought', 'were', 'my', \"wouldn't\", \"we'd\", 'and', 'then', 'is', \"didn't\", 'few', 'an', 'as', 'itself', 'at', 'have', 'in', 'any', 'if', 'these', \"let's\", 'no', 'that', 'when', 'same', 'how', 'other', 'which', 'you', \"shan't\", 'again', 'our', 'who', \"where's\", 'most', 'such', 'on', 'why', 'a', 'off', 'i', \"she'd\", 'having', \"you'll\", 'so', \"we're\", \"she's\", 'the', 'once', 'yours', \"that's\"]\n"
     ]
    }
   ],
   "source": [
    "print list(stop_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sentences = tokenize(d_train.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from gensim.models.word2vec import Word2Vec\n",
    "model = Word2Vec(sentences, size=300, window=10, min_count=1, sample=1e-3, workers=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model.init_sims(replace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  3.68031524e-02,  -1.73313618e-02,   2.39172988e-02,\n",
       "         2.63126623e-02,  -2.09560362e-03,   2.88584847e-02,\n",
       "         1.84983388e-02,   1.61847845e-02,   3.55088450e-02,\n",
       "        -1.09551460e-01,   8.66681430e-03,   5.40465722e-03,\n",
       "         3.01059918e-04,  -6.86682537e-02,   9.07223746e-02,\n",
       "        -2.10372563e-02,   3.71289365e-02,   7.52746388e-02,\n",
       "         2.14488988e-04,  -1.00362170e-02,  -1.01014741e-01,\n",
       "         1.02308214e-01,   3.20761763e-02,  -5.46756908e-02,\n",
       "        -1.42806098e-02,   5.29203713e-02,   6.76437244e-02,\n",
       "         4.52545471e-02,  -3.38600017e-02,  -9.30002108e-02,\n",
       "         2.45192088e-02,   5.53836077e-02,  -7.16699800e-03,\n",
       "         5.90483360e-02,  -8.47217366e-02,   6.08438440e-02,\n",
       "        -7.18232915e-02,   4.67804708e-02,   1.32195026e-01,\n",
       "         1.17428429e-01,  -4.01911624e-02,   6.89821271e-03,\n",
       "         2.51749437e-02,   2.66396683e-02,   3.70586081e-03,\n",
       "         1.63978413e-02,   4.36227731e-02,  -7.04589635e-02,\n",
       "        -1.55292740e-02,   8.42350423e-02,   1.51802972e-01,\n",
       "         3.18416655e-02,  -4.92057689e-02,  -3.66481440e-03,\n",
       "        -4.09203470e-02,  -1.18032508e-02,   9.20815486e-03,\n",
       "         6.21036775e-02,  -6.40419051e-02,  -4.64000776e-02,\n",
       "         1.51039407e-01,  -1.50426120e-01,  -1.33195787e-03,\n",
       "        -1.60630178e-02,  -6.76299818e-03,  -4.00880203e-02,\n",
       "        -4.95725572e-02,   2.77464017e-02,  -5.78683876e-02,\n",
       "        -4.15379368e-03,  -6.82369387e-03,  -1.12992570e-01,\n",
       "         1.90758295e-02,  -3.58685909e-04,  -8.61261711e-02,\n",
       "         6.88697174e-02,  -7.72073492e-02,   5.14152534e-02,\n",
       "         4.87889685e-02,  -9.20623261e-03,   2.75032986e-02,\n",
       "        -8.93953349e-03,  -7.17302263e-02,   2.63757426e-02,\n",
       "         1.17861174e-01,  -9.01978835e-02,  -9.88838151e-02,\n",
       "        -1.53771825e-02,  -6.96591660e-02,  -1.07649993e-03,\n",
       "        -1.11715309e-01,  -5.69908284e-02,  -3.45526747e-02,\n",
       "        -1.03636552e-02,  -1.06374146e-02,  -4.84549142e-02,\n",
       "        -5.06430119e-02,  -9.12421271e-02,  -4.20696139e-02,\n",
       "         7.88904428e-02,  -6.89790249e-02,   7.52062025e-03,\n",
       "        -2.90804580e-02,   3.07238027e-02,   6.54164553e-02,\n",
       "         3.93598229e-02,   1.45371864e-02,   6.26129424e-03,\n",
       "        -9.42131132e-02,   8.22537392e-03,  -7.89110214e-02,\n",
       "         8.79941583e-02,   1.90700181e-02,  -1.34457862e-02,\n",
       "        -4.17764559e-02,   1.03315689e-01,   3.10423244e-02,\n",
       "        -1.09508671e-02,   1.61643643e-02,   8.63924101e-02,\n",
       "         9.44718905e-03,  -4.84026298e-02,   3.16865556e-02,\n",
       "        -3.33051052e-04,  -8.11982155e-02,  -4.91626225e-02,\n",
       "         3.83723192e-02,   9.36327651e-02,   1.75947472e-02,\n",
       "         7.99207482e-03,  -8.33873153e-02,   7.32554644e-02,\n",
       "         8.55481252e-02,   3.38279977e-02,  -6.30003661e-02,\n",
       "        -1.06950305e-01,  -2.57532764e-03,  -8.69107619e-02,\n",
       "        -8.38905945e-02,  -2.52390001e-02,   4.47176024e-02,\n",
       "         2.98092533e-02,  -3.36933024e-02,   5.62217310e-02,\n",
       "         4.98724449e-03,  -1.04179740e-01,   4.71587963e-02,\n",
       "         8.17830581e-03,  -1.81893595e-02,  -6.29652515e-02,\n",
       "         2.12100241e-03,  -1.26776434e-02,  -3.99298556e-02,\n",
       "        -7.92319179e-02,  -3.71179506e-02,  -4.98937219e-02,\n",
       "        -4.40028273e-02,  -6.31705299e-02,   2.12799329e-02,\n",
       "         1.09728582e-01,  -3.11787007e-04,   8.88880417e-02,\n",
       "         4.32723388e-02,  -9.17474031e-02,   7.68122748e-02,\n",
       "        -9.60509703e-02,   1.79586820e-02,   1.54920649e-02,\n",
       "        -5.79718761e-02,  -9.94104985e-03,  -6.22458830e-02,\n",
       "        -5.81062352e-03,   3.41688655e-02,  -3.06482129e-02,\n",
       "         8.93115476e-02,   1.56571844e-03,  -3.13598663e-02,\n",
       "         2.58147363e-02,   8.44683573e-02,  -7.09150508e-02,\n",
       "         3.52193490e-02,  -8.44642222e-02,   1.15281150e-01,\n",
       "         4.13945317e-02,  -7.53380284e-02,   1.06682628e-02,\n",
       "        -2.90197339e-02,  -1.93742830e-02,   2.80632749e-02,\n",
       "         5.98113611e-03,  -4.81246673e-02,   5.81961051e-02,\n",
       "        -1.21646203e-01,  -7.56746012e-05,   1.37762520e-02,\n",
       "        -1.07660465e-01,  -5.49132526e-02,  -2.08277833e-02,\n",
       "        -1.07252866e-01,   7.81668052e-02,  -1.25467703e-01,\n",
       "         1.28760673e-02,   3.40522267e-02,  -4.53600958e-02,\n",
       "         3.00191492e-02,  -1.05644763e-03,   2.28488427e-02,\n",
       "        -1.36948330e-02,  -4.83282395e-02,   3.76505554e-02,\n",
       "         2.14589084e-03,  -6.56597763e-02,  -2.17904039e-02,\n",
       "        -1.30352750e-01,   1.00631500e-02,  -2.68849730e-02,\n",
       "         3.97608057e-02,  -7.62850717e-02,  -1.71926096e-02,\n",
       "         9.73487180e-03,   1.00726984e-01,  -3.03553939e-02,\n",
       "         3.96722928e-02,  -3.65546495e-02,  -4.54760306e-02,\n",
       "        -1.08036343e-02,  -2.10739505e-02,  -1.74224488e-02,\n",
       "         2.45836675e-02,   9.26169008e-02,   1.69083904e-02,\n",
       "        -1.93465408e-02,   4.19885060e-03,   6.44169822e-02,\n",
       "         2.11614668e-02,   7.29327044e-03,  -1.25794038e-01,\n",
       "         2.51813382e-02,   7.73636624e-02,   8.36196244e-02,\n",
       "        -4.76672724e-02,   3.01581249e-03,  -1.89450905e-02,\n",
       "        -2.14905385e-02,   7.21007884e-02,  -5.47636338e-02,\n",
       "         7.87780955e-02,   6.38814420e-02,  -1.82743594e-02,\n",
       "         2.52366532e-03,   5.32709819e-04,  -1.36896092e-02,\n",
       "         7.07334876e-02,  -6.54596016e-02,  -9.79023874e-02,\n",
       "         4.26013544e-02,   6.34806380e-02,   1.94339901e-02,\n",
       "        -1.79679878e-02,   3.58978361e-02,  -2.59635057e-02,\n",
       "        -5.14540309e-03,  -3.97503264e-02,   3.99754904e-02,\n",
       "         1.13015451e-01,   1.92152523e-02,   8.84376913e-02,\n",
       "        -6.68382198e-02,   1.51651455e-02,   1.41883986e-02,\n",
       "         3.25143524e-02,   4.63566855e-02,  -1.13264881e-01,\n",
       "        -1.22436136e-02,  -3.99175175e-02,  -1.39550604e-02,\n",
       "        -6.48894459e-02,  -9.74341184e-02,  -1.99261121e-02,\n",
       "         1.34053323e-02,  -1.10854441e-02,  -2.31818724e-02,\n",
       "         6.56687096e-02,   9.27144662e-02,   6.97587132e-02,\n",
       "         3.42045613e-02,   4.58224751e-02,  -1.15564905e-01,\n",
       "        -4.97712307e-02,  -1.05411328e-01,  -5.43938540e-02,\n",
       "        -1.27959177e-02,  -1.02667041e-01,  -9.49689746e-03,\n",
       "         3.08882631e-02,   1.99520700e-02,  -5.47346135e-04,\n",
       "         5.61500862e-02,  -4.44469228e-02,  -2.57499572e-02], dtype=float32)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model['movie']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def featurize_w2v(model, sentences):\n",
    "    f = zeros((len(sentences), model.vector_size))\n",
    "    for i,s in enumerate(sentences):\n",
    "        for w in s:\n",
    "            try:\n",
    "                vec = model[w]\n",
    "            except KeyError:\n",
    "                continue\n",
    "            f[i,:] = f[i,:] + vec\n",
    "        f[i,:] = f[i,:] / len(s)\n",
    "    return f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "features_w2v = featurize_w2v(model, sentences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=-1,\n",
       "            oob_score=False, random_state=None, verbose=0,\n",
       "            warm_start=False)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model4 = RandomForestClassifier(n_estimators=100, n_jobs=-1)\n",
    "model4.fit(features_w2v, d_train.sentiment)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_sentences = tokenize(d_test.review)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_features_w2v = featurize_w2v(model, test_sentences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pred4 = model4.predict_proba(test_features_w2v)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.6, 1)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEMCAYAAAArnKpYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngiAIIuf/A2Zj9y4g9NaHAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12f0e4590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "performance(d_test.sentiment, pred1, ann=False)\n",
    "performance(d_test.sentiment, pred2, color=\"b\", ann=False)\n",
    "performance(d_test.sentiment, pred3, color=\"r\", ann=False)\n",
    "performance(d_test.sentiment, pred4, color=\"c\")\n",
    "xlim(0,0.3)\n",
    "ylim(0.6,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 0, 0, 0])"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "examples = [\n",
    "        \"This movie is bad\",\n",
    "        \"This movie is great\",\n",
    "        \"I was going to say something awesome, but I simply can't because the movie is so bad.\",\n",
    "        \"I was going to say something awesome or great or good, but I simply can't because the movie is so bad.\",\n",
    "        \"It might have bad actors, but everything else is good.\"\n",
    "    ]\n",
    "example_feat4 = featurize_w2v(model, tokenize(examples))\n",
    "model4.predict(example_feat4)"
   ]
  }
 ],
 "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": 0
}