exploration.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv(\"data/labeledTrainData.tsv\", header=0, delimiter=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from bs4 import BeautifulSoup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_html(row):\n",
    "    return BeautifulSoup(row['review'], 'html.parser').get_text()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data['review'] = data.apply(remove_html, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.tokenize import TweetTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "tweeter_tokenizer = TweetTokenizer()\n",
    "import nltk\n",
    "vectorizer = TfidfVectorizer(tokenizer=nltk.word_tokenize, stop_words='english', max_features=5000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = vectorizer.fit_transform(data['review'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25000, 1000)"
      ]
     },
     "execution_count": 10,
     "output_type": "execute_result",
     "metadata": {}
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "forest = RandomForestClassifier(n_estimators = 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "forest = forest.fit( X_train.toarray(), data[\"sentiment\"] )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "test = pd.read_csv(\"data/testData.tsv\", header=0, delimiter=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "test['review'] = test.apply(remove_html, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test = vectorizer.fit_transform(test['review'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = forest.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_confusion_matrix(cm, title='Confusion matrix', cmap=plt.cm.Blues):\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(set(sentiments)))\n",
    "    plt.xticks(tick_marks, set(sentiments), rotation=45)\n",
    "    plt.yticks(tick_marks, set(sentiments))\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.cross_validation import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "reviews = data['review']\n",
    "sentiments = data['sentiment']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(reviews, sentiments, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = vectorizer.fit_transform(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test = vectorizer.fit_transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "forest = RandomForestClassifier(n_estimators = 100)\n",
    "forest = forest.fit( X_train.toarray(), y_train )\n",
    "predictions = forest.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVMAAAEpCAYAAAAu8sHTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFcBhamZWgP8POTbik6AHRo0AAAAASUVORK5CYII=\n"
     },
     "output_type": "display_data",
     "metadata": {}
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVAAAAEpCAYAAADFxXrQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQM3MSnICNTMryQnUzKwkJ1Azs5L+P+RjxCZOwyxoAAAAAElFTkSuQmCC\n"
     },
     "output_type": "display_data",
     "metadata": {}
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, predictions)\n",
    "np.set_printoptions(precision=2)\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cm)    \n",
    "\n",
    "cm_normalized = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cm_normalized, title='Normalized confusion matrix')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve, auc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEZCAYAAACNebLAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY4wxIVmCMMYYE9L/A5lPuBzaBLaVAAAAAElFTkSuQmCC\n"
     },
     "output_type": "display_data",
     "metadata": {}
    }
   ],
   "source": [
    "false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test, predictions)\n",
    "roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "plt.plot(false_positive_rate, true_positive_rate, 'b', label='%s: AUC %0.2f'% ('model',roc_auc))\n",
    "plt.title('Classifiers comparaison with ROC')\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.xlim([-0.1,1.2])\n",
    "plt.ylim([-0.1,1.2])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "   positive       0.58      0.57      0.57      2481\n",
      "   negative       0.58      0.59      0.58      2519\n",
      "\n",
      "avg / total       0.58      0.58      0.58      5000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_test, predictions, target_names = [\"positive\", \"negative\"]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    ""
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python3",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3.0
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}