L7 - Multinomial and Bernoulli Naive Bayes Classifiers.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predict Spam SMS using Naive Bayes Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.naive_bayes import MultinomialNB,BernoulliNB\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.feature_extraction import DictVectorizer\n",
    "from sklearn.cross_validation import KFold\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use [SMS Spam Collection Data Set](https://archive.ics.uci.edu/ml/datasets/SMS+Spam+Collection) from UCI, which contains 5574 SMS, 747 of them are spam and the rest are ham."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>category</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ham</td>\n",
       "      <td>Go until jurong point, crazy.. Available only in bugis n great world la e buffet... Cine there got amore wat...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>Ok lar... Joking wif u oni...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>spam</td>\n",
       "      <td>Free entry in 2 a wkly comp to win FA Cup final tkts 21st May 2005. Text FA to 87121 to receive entry question(std txt rate)T&amp;C's apply 08452810075over18's</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>U dun say so early hor... U c already then say...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ham</td>\n",
       "      <td>Nah I don't think he goes to usf, he lives around here though</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  category  \\\n",
       "0      ham   \n",
       "1      ham   \n",
       "2     spam   \n",
       "3      ham   \n",
       "4      ham   \n",
       "\n",
       "                                                                                                                                                          text  \n",
       "0                                              Go until jurong point, crazy.. Available only in bugis n great world la e buffet... Cine there got amore wat...  \n",
       "1                                                                                                                                Ok lar... Joking wif u oni...  \n",
       "2  Free entry in 2 a wkly comp to win FA Cup final tkts 21st May 2005. Text FA to 87121 to receive entry question(std txt rate)T&C's apply 08452810075over18's  \n",
       "3                                                                                                            U dun say so early hor... U c already then say...  \n",
       "4                                                                                                Nah I don't think he goes to usf, he lives around here though  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('./asset/SMSSpamCollection.txt', sep='\\t')\n",
    "pd.set_option(\"display.max_colwidth\",999)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use the count of each tokens (words) in the SMS as its feature. \n",
    "\n",
    "Firstly we convert the SMS into features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'e': 1, 'got': 1, 'bugis': 1, 'great': 1, 'world': 1, 'in': 1, 'Available': 1, 'Go': 1, 'jurong': 1, 'there': 1, 'crazy..': 1, 'buffet...': 1, 'la': 1, 'wat...': 1, 'amore': 1, 'until': 1, 'Cine': 1, 'only': 1, 'n': 1, 'point,': 1}\n"
     ]
    }
   ],
   "source": [
    "def get_freq_of_tokens(sms):\n",
    "    tokens = {}\n",
    "    for token in sms.split(' '):\n",
    "        if token not in tokens:\n",
    "            tokens[token] = 1\n",
    "        else:\n",
    "            tokens[token] += 1\n",
    "    return tokens\n",
    "\n",
    "print(get_freq_of_tokens(data.iloc[0].text))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "features_and_labels = []\n",
    "for index in range(len(data)):\n",
    "    features_and_labels.append((get_freq_of_tokens(data.iloc[index].text), data.iloc[index].category))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We then trainsform the features into a sparse martix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<5572x15693 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 81504 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "encoder = LabelEncoder()\n",
    "vectorizer = DictVectorizer(dtype=float, sparse=True)\n",
    "X, y = list(zip(*features_and_labels))\n",
    "X = vectorizer.fit_transform(X)\n",
    "    # print(X)\n",
    "    # print(vectorizer.feature_names_)\n",
    "y = encoder.fit_transform(y)\n",
    "\n",
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use the first SMS (i.e., X[0]) to show how to get the feature names and values. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Available:1.0\n",
      "Cine:1.0\n",
      "Go:1.0\n",
      "amore:1.0\n",
      "buffet...:1.0\n",
      "bugis:1.0\n",
      "crazy..:1.0\n",
      "e:1.0\n",
      "got:1.0\n",
      "great:1.0\n",
      "in:1.0\n",
      "jurong:1.0\n",
      "la:1.0\n",
      "n:1.0\n",
      "only:1.0\n",
      "point,:1.0\n",
      "there:1.0\n",
      "until:1.0\n",
      "wat...:1.0\n",
      "world:1.0\n"
     ]
    }
   ],
   "source": [
    "arr = X[0].toarray()\n",
    "for i in range(len(arr[0])):\n",
    "    if arr[0][i] > 0:\n",
    "        print('{}:{}'.format(vectorizer.feature_names_[i], arr[0][i]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build Multinomial Naive Bayes Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`alpha` is the additive (Laplace/Lidstone) smoothing parameter. Its default value is 1.0 -- we will see that this works pretty well for our problem later. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.99246231155778897"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb = MultinomialNB(alpha=1)\n",
    "nb.fit(X, y)\n",
    "nb.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's use cross validation to explore how alpha affects the performance of the classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "n_folds = 10\n",
    "kf = KFold(n=X.shape[0], n_folds=n_folds, shuffle=True, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def test_Multinomial_NB(train_X, train_y, test_X, test_y, alpha=1, debug = False):\n",
    "    nb = MultinomialNB(alpha=alpha)\n",
    "    nb.fit(train_X, train_y)\n",
    "    train_error = nb.score(train_X, train_y)\n",
    "    test_error = nb.score(test_X, test_y)\n",
    "    if debug:\n",
    "        print('training error:\\t{}'.format(train_error))\n",
    "        print('testing error:\\t{}'.format(test_error))\n",
    "    return train_error, test_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def cv_MultinomialNB(alpha = 1.0):\n",
    "    train_error_total = 0\n",
    "    test_error_total = 0\n",
    "    for train, test in kf:\n",
    "        train_X = X[train]\n",
    "        test_X = X[test]\n",
    "        train_y = y[train]\n",
    "        test_y = y[test]\n",
    "        train_error, test_error = test_Multinomial_NB(train_X, train_y, test_X, test_y, alpha)\n",
    "        train_error_total += train_error\n",
    "        test_error_total += test_error\n",
    "    return train_error_total/n_folds, test_error_total/n_folds\n",
    "#     print('===================')\n",
    "#     print('avg. training error:\\t{}'.format(train_error_total/n_folds))\n",
    "#     print('avg. testing error:\\t{}'.format(test_error_total/n_folds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def cv_plot_MultinomialNB():\n",
    "    cv_res = []\n",
    "    rng = sorted([0.5] + list(range(0, 10)))\n",
    "    for i in rng:\n",
    "        train_error, test_error = cv_MultinomialNB(i)\n",
    "        cv_res.append([i, train_error, test_error])\n",
    "    cv_res_arr = np.array(cv_res)\n",
    "    plt.figure(figsize=(16,9)) \n",
    "    plt.title('Error vs. alpha') \n",
    "    plot_train, = plt.plot(cv_res_arr[:,0], cv_res_arr[:,1], label='training')\n",
    "    plot_test, = plt.plot(cv_res_arr[:,0], cv_res_arr[:,2], label='testing')\n",
    "    plt.legend(handles=[plot_train, plot_test])\n",
    "    plt.ylim((min(min(cv_res_arr[:,1]), min(cv_res_arr[:,2])) - 0.01, max(max(cv_res_arr[:,1]), max(cv_res_arr[:,2]))+0.01))\n",
    "    plt.xticks(rng)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA60AAAIhCAYAAACloFzIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a23786b470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cv_plot_MultinomialNB()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Bernoulli Naive Bayes Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that since we have the count matrix for words in the vocabulary, we can binarize it to be used in a Bernoulli NB classifier. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.98438621679827709"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nb = BernoulliNB(alpha=1, binarize=0.0) # every value >0.0 will be binarized to 1\n",
    "nb.fit(X, y)\n",
    "nb.score(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def test_Bernoulli_NB(train_X, train_y, test_X, test_y, alpha=1, debug = False):\n",
    "    nb = BernoulliNB(alpha=alpha)\n",
    "    nb.fit(train_X, train_y)\n",
    "    train_error = nb.score(train_X, train_y)\n",
    "    test_error = nb.score(test_X, test_y)\n",
    "    if debug:\n",
    "        print('training error:\\t{}'.format(train_error))\n",
    "        print('testing error:\\t{}'.format(test_error))\n",
    "    return train_error, test_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def cv_BernoulliNB(alpha = 1.0):\n",
    "    train_error_total = 0\n",
    "    test_error_total = 0\n",
    "    for train, test in kf:\n",
    "        train_X = X[train]\n",
    "        test_X = X[test]\n",
    "        train_y = y[train]\n",
    "        test_y = y[test]\n",
    "        train_error, test_error = test_Bernoulli_NB(train_X, train_y, test_X, test_y, alpha)\n",
    "        train_error_total += train_error\n",
    "        test_error_total += test_error\n",
    "    return train_error_total/n_folds, test_error_total/n_folds\n",
    "#     print('===================')\n",
    "#     print('avg. training error:\\t{}'.format(train_error_total/n_folds))\n",
    "#     print('avg. testing error:\\t{}'.format(test_error_total/n_folds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def cv_plot_BernoulliNB():\n",
    "    cv_res = []\n",
    "    rng = sorted([0.5] + list(range(0, 10)))\n",
    "    for i in rng:\n",
    "        train_error, test_error = cv_BernoulliNB(i)\n",
    "        cv_res.append([i, train_error, test_error])\n",
    "    cv_res_arr = np.array(cv_res)\n",
    "    plt.figure(figsize=(16,9)) \n",
    "    plt.title('Error vs. alpha')\n",
    "    plot_train, = plt.plot(cv_res_arr[:,0], cv_res_arr[:,1], label='training')\n",
    "    plot_test, = plt.plot(cv_res_arr[:,0], cv_res_arr[:,2], label='testing')\n",
    "    plt.legend(handles=[plot_train, plot_test])\n",
    "    plt.ylim((min(min(cv_res_arr[:,1]), min(cv_res_arr[:,2])) - 0.01, max(max(cv_res_arr[:,1]), max(cv_res_arr[:,2]))+0.01))\n",
    "    plt.xticks(rng)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA60AAAIhCAYAAACloFzIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a23797c358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cv_plot_BernoulliNB()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}