06 - Working With Text Data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib notebook\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Text Classification of Movie Reviews"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Unpack data - this only works on linux and (maybe?) OS X. Unpack using 7zip on Windows."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#! tar -xf data/aclImdb.tar.bz2 --directory data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_files\n",
    "\n",
    "reviews_train = load_files(\"data/aclImdb/train/\")\n",
    "text_train, y_train = reviews_train.data, reviews_train.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print(\"Number of documents in training data: %d\" % len(text_train))\n",
    "print(np.bincount(y_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "reviews_test = load_files(\"data/aclImdb/test/\")\n",
    "text_test, y_test = reviews_test.data, reviews_test.target\n",
    "print(\"Number of documents in test data: %d\" % len(text_test))\n",
    "print(np.bincount(y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Subsample for interactivity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "text_train, y_train = text_train[::10], y_train[::10]\n",
    "text_test, y_test = text_test[::10], y_test[::10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from IPython.display import HTML"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print(text_train[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "HTML(text_train[1].decode(\"utf-8\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print(y_train[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "cv = CountVectorizer()\n",
    "cv.fit(text_train)\n",
    "\n",
    "len(cv.vocabulary_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "print(cv.get_feature_names()[:50])\n",
    "print(cv.get_feature_names()[5000:5020])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X_train = cv.transform(text_train)\n",
    "X_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print(text_train[739])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X_train[739].nonzero()[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X_test = cv.transform(text_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.svm import LinearSVC\n",
    "\n",
    "svm = LinearSVC()\n",
    "svm.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "svm.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "svm.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def visualize_coefficients(classifier, feature_names, n_top_features=25):\n",
    "    # get coefficients with large absolute values \n",
    "    coef = classifier.coef_.ravel()\n",
    "    positive_coefficients = np.argsort(coef)[-n_top_features:]\n",
    "    negative_coefficients = np.argsort(coef)[:n_top_features]\n",
    "    interesting_coefficients = np.hstack([negative_coefficients, positive_coefficients])\n",
    "    # plot them\n",
    "    plt.figure(figsize=(15, 5))\n",
    "    colors = [\"red\" if c < 0 else \"blue\" for c in coef[interesting_coefficients]]\n",
    "    plt.bar(np.arange(2 * n_top_features), coef[interesting_coefficients], color=colors)\n",
    "    feature_names = np.array(feature_names)\n",
    "    plt.subplots_adjust(bottom=0.3)\n",
    "    plt.xticks(np.arange(1, 1 + 2 * n_top_features), feature_names[interesting_coefficients], rotation=60, ha=\"right\");\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "visualize_coefficients(svm, cv.get_feature_names())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "cv = CountVectorizer(ngram_range=(1, 2))\n",
    "cv.fit(text_train)\n",
    "len(cv.vocabulary_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_test = cv.transform(text_test)\n",
    "X_train = cv.transform(text_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "svm = LinearSVC()\n",
    "svm.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "visualize_coefficients(svm, cv.get_feature_names())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Look at SpaCy and NLTK"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exercises\n",
    "* Create a pipeine using the count vectorizer and SVM (see 07). Train and score using the pipeline.\n",
    "* vizualize the coefficients of the svm inside the pipeline using pipeline.named_steps['linearsvc'] and pipeline.named_steps['countvectorizer']\n",
    "* Grid search the C in the LinearSVC using the pipeline.\n",
    "\n",
    "* Vary the n_gram_range in the count vectorizer, visualize the changed coefficients.\n",
    "* Grid search the C in the LinearSVC together with the n_gram_range (try (1,1), (1, 2), (2, 2))\n",
    "* Visualize the parameter again for the grid-searched model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# %load solutions/text_pipeline.py"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.4.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}