AmazonFineFoodReviews.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sentimentic analysis on Amazon Fine Food Reviews Dataset using NLTK methodology, naive Bayes, Logistic Regression and metrics like roc_curve and auc value, confusion_matrix and and classification_report"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "The purpose of this analysis is to make up a prediction model where we will be able to predict whether a recommendation is positive or negative. In this analysis, we will not focus on the Score, but only the positive/negative sentiment of the recommendation.\n",
    "In order to load the data, I will make use of sqlite3 package where only Score and Summary columns are taken nto th consideration. As we only want to fiqure out the sentiment analysis( positive or negative)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Score</th>\n",
       "      <th>Summary</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5</td>\n",
       "      <td>Good Quality Dog Food</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>Not as Advertised</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4</td>\n",
       "      <td>\"Delight\" says it all</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>Cough Medicine</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>Great taffy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Score                Summary\n",
       "0      5  Good Quality Dog Food\n",
       "1      1      Not as Advertised\n",
       "2      4  \"Delight\" says it all\n",
       "3      2         Cough Medicine\n",
       "4      5            Great taffy"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import sqlite3\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import string\n",
    "import nltk\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "from sklearn.cross_validation import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from nltk.stem.porter import PorterStemmer\n",
    "\n",
    "con=sqlite3.connect('database.sqlite')\n",
    "messages=pd.read_sql_query(\"\"\" SELECT Score, Summary from Reviews where Score!=3\"\"\", con)\n",
    "messages.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Changing the score into positive and negative recommendation with a score < 3 goes in the negative sentiment list and score >= 3 goes in the positive sentiment list."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "def partition(x):\n",
    "    if x < 3:\n",
    "        return 'negative'\n",
    "    return 'positive'\n",
    "\n",
    "Score=messages['Score']\n",
    "Score= Score.map(partition)\n",
    "Summary=messages['Summary']\n",
    "X_train, X_test, y_train, y_test = train_test_split(Summary, Score, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Score</th>\n",
       "      <th>Summary</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>positive</td>\n",
       "      <td>Good Quality Dog Food</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>negative</td>\n",
       "      <td>Not as Advertised</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>positive</td>\n",
       "      <td>\"Delight\" says it all</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>negative</td>\n",
       "      <td>Cough Medicine</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>positive</td>\n",
       "      <td>Great taffy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Score                Summary\n",
       "0  positive  Good Quality Dog Food\n",
       "1  negative      Not as Advertised\n",
       "2  positive  \"Delight\" says it all\n",
       "3  negative         Cough Medicine\n",
       "4  positive            Great taffy"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp=messages\n",
    "tmp['Score']=tmp['Score'].map(partition)\n",
    "tmp.head()"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "Defining and making use of some nltk packages which include PorterStemmer(), word_tokenize().\n",
    "Making use of sting package to remove punctuation from text data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "stemmer=PorterStemmer()\n",
    "def stem_tokens(token, stemmer):\n",
    "    stemmed=[]\n",
    "    for item in token:\n",
    "        stemmed.append(stemmer.stem(item))\n",
    "    return stemmed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "def tokenize(text):\n",
    "    token=nltk.word_tokenize(text)\n",
    "    stems=stem_tokens(token,stemmer)\n",
    "    return ' '.join(stems)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "intab = string.punctuation\n",
    "outtab = \"                                \"\n",
    "trantab = str.maketrans(intab, outtab)\n"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "Different processes are applied on text in X_train and X_test data \n",
    "Changing the upper case letter to the lower case.\n",
    "Making use of maketrans to remove punctuation.\n",
    "Tokenizing the text data with nltk's word_tokenize package.\n",
    "Finally, when the text data is ready, applying CountVectorizer() and Tfidf_transformer() on them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "corpus=[]\n",
    "for text in X_train:\n",
    "    text=text.lower()\n",
    "    text=text.translate(trantab)\n",
    "    text=tokenize(text)\n",
    "    corpus.append(text)\n",
    "    \n",
    "count_vect = CountVectorizer()\n",
    "X_train_counts = count_vect.fit_transform(corpus)        \n",
    "        \n",
    "tfidf_transformer = TfidfTransformer()\n",
    "X_train_tfidf = tfidf_transformer.fit_transform(X_train_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Same procedure is applied on the test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true,
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [],
   "source": [
    "test_set=[]\n",
    "for text in X_test:\n",
    "    text=text.lower()\n",
    "    text=text.translate(trantab)\n",
    "    text=tokenize(text)\n",
    "    test_set.append(text)\n",
    "    \n",
    "X_new_counts = count_vect.transform(test_set)\n",
    "X_test_tfidf = tfidf_transformer.transform(X_new_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The changes before and after all the preprocess and nltk techniques."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "slideshow": {
     "slide_type": "subslide"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>After</th>\n",
       "      <th>Before</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>496497</th>\n",
       "      <td>almond great buy</td>\n",
       "      <td>ALMONDS GREAT BUY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>225396</th>\n",
       "      <td>i never thought i d have to say no to more fru...</td>\n",
       "      <td>I never thought i'd have to say no to more fru...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>288197</th>\n",
       "      <td>we love thi stuff</td>\n",
       "      <td>We love this Stuff</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88450</th>\n",
       "      <td>fan friggen tastic</td>\n",
       "      <td>Fan-friggen-tastic</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>354669</th>\n",
       "      <td>great for offic</td>\n",
       "      <td>Great for office</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                    After  \\\n",
       "496497                                   almond great buy   \n",
       "225396  i never thought i d have to say no to more fru...   \n",
       "288197                                  we love thi stuff   \n",
       "88450                                  fan friggen tastic   \n",
       "354669                                    great for offic   \n",
       "\n",
       "                                                   Before  \n",
       "496497                                  ALMONDS GREAT BUY  \n",
       "225396  I never thought i'd have to say no to more fru...  \n",
       "288197                                 We love this Stuff  \n",
       "88450                                  Fan-friggen-tastic  \n",
       "354669                                   Great for office  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df=pd.DataFrame({'Before': X_train, 'After': corpus})\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {
    "slideshow": {
     "slide_type": "notes"
    }
   },
   "source": [
    "Defining a dictionary predictors which will contain the predicted score for all the rows and for all the machine learning models.\n",
    "Applying MultinomialNB from sklearn.naive_bayes:\n",
    "How MultinomialNB works: \"it counts how often word occurs in the data\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "predictors={}\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "model=MultinomialNB().fit(X_train_tfidf, y_train)\n",
    "predictors['Multinomial']= model.predict(X_test_tfidf)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "Applying BernoulliNB from sklearn.naive_bayes:\n",
    "Here Bernoulli would be applied as a text classification with bag of words model where this model checks where that \"specific word occurred in the document\" or not."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import BernoulliNB\n",
    "model=BernoulliNB().fit(X_train_tfidf, y_train)\n",
    "predictors['Bernoulli']= model.predict(X_test_tfidf)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Applying LogisticRegression:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn import linear_model\n",
    "logreg = linear_model.LogisticRegression(C=1e6)\n",
    "logreg.fit(X_train_tfidf, y_train)\n",
    "predictors['Logistic'] = logreg.predict(X_test_tfidf)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {},
   "source": [
    "To evaluate the performance of above models, I'm making use of roc_curve and auc from sklearn.metrics package.\n",
    "With some exciting use of matplotlib library, I can actually represent this metrics model with a nice plot as shown in the output down.\n",
    "The curve with highest AUC value will show our best algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26d0bd1a518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def formatt(x):\n",
    "    if x == 'negative':\n",
    "        return 0\n",
    "    return 1\n",
    "vfunc = np.vectorize(formatt)\n",
    "\n",
    "cmp = 0\n",
    "colors = ['b', 'g', 'y', 'm', 'k']\n",
    "for model, predicted in predictors.items():\n",
    "    false_positive_rate, true_positive_rate, thresholds = roc_curve(y_test.map(formatt), vfunc(predicted))\n",
    "    roc_auc = auc(false_positive_rate, true_positive_rate)\n",
    "    plt.plot(false_positive_rate, true_positive_rate, colors[cmp], label='%s: AUC %0.2f'% (model,roc_auc))\n",
    "    cmp += 1\n",
    "\n",
    "plt.title('Classifiers comparaison with ROC')\n",
    "plt.legend(loc='lower right')\n",
    "plt.plot([0,1],[0,1],'r--')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Understanding the model result with classification_report and confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "   negative       0.80      0.69      0.74     16379\n",
      "   positive       0.94      0.97      0.96     88784\n",
      "\n",
      "avg / total       0.92      0.93      0.92    105163\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, predictors['Logistic']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWIAAAEmCAYAAABYlZoAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkjkQm5mV7P8DJZzeh8EN4vYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26d00a5b6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVIAAAEmCAYAAAAwZhg4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x26d013d9668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "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(Score)))\n",
    "    plt.xticks(tick_marks, set(Score), rotation=45)\n",
    "    plt.yticks(tick_marks, set(Score))\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    \n",
    "# Compute confusion matrix\n",
    "cm = confusion_matrix(y_test, predictors['Logistic'])\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": 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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}