Added python files for rating prediction using TFIDF and LDA+sentiment

Author: pragyaarora

rating_prediction_lda.py

@@ -194,7 +194,10 @@ def getTopicDistMatrix(lda, totalTopics, corpus, all_dist, star):
 print "Number of 2-star reviews after processing: ", len(corpus_2stars_test)
 print "Number of 1-star reviews after processing: ", len(corpus_1stars_test)
 
-
+all_5_4_test = np.append(corpus_5stars_test, corpus_4stars_test)
+all_5_4_3_test = np.append(all_5_4_test, corpus_3stars_test)
+all_5_4_3_2_test = np.append(all_5_4_3_test, corpus_2stars_test)
+all_text_test = np.append(all_5_4_3_2_test, corpus_1stars_test)
 
 topic_dist_list = []
 

rating_prediction_lda_sentiment.py

@@ -0,0 +1,86 @@
+from sklearn import metrics
+from sklearn.metrics import classification_report
+from sklearn.naive_bayes import MultinomialNB
+from sklearn.linear_model import LogisticRegression
+from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier
+from sklearn.metrics import accuracy_score
+
+from rating_prediction_lda import totalTopics,all_text_train, all_text_test,topic_dist_train_all_stars,topic_dist_test_all_stars
+from rating_prediction_tfidf import tfidfvectorizer
+
+def getSentiment(s):
+    if s < 3.5:
+        return 0
+    else:
+        return 1
+
+
+topic_dist_train_all_stars['Sentiment'] = topic_dist_train_all_stars['Star'].map(getSentiment)
+topic_dist_test_all_stars['Sentiment'] = topic_dist_test_all_stars['Star'].map(getSentiment)
+
+sentimentTextTrain = tfidfvectorizer.fit_transform(all_text_train)
+sentimentTextTest = tfidfvectorizer.transform(all_text_test)
+
+sentimentLabelTrain = topic_dist_train_all_stars['Sentiment']
+sentimentLabelTest = topic_dist_test_all_stars['Sentiment']
+
+classifier = LogisticRegression().fit(sentimentTextTrain, sentimentLabelTrain)
+
+ySentimentTrain = classifier.predict(sentimentTextTrain)
+ySentimentTest = classifier.predict(sentimentTextTest)
+
+topic_dist_train_all_stars['Sentiment_Predicted'] = ySentimentTrain
+topic_dist_test_all_stars['Sentiment_Predicted'] = ySentimentTest
+
+
+features = list(topic_dist_train_all_stars.columns[:totalTopics])
+features.append(topic_dist_train_all_stars.columns[totalTopics+2])
+
+
+x_train = topic_dist_train_all_stars[features]
+y_train = topic_dist_train_all_stars['Star']
+
+x_test = topic_dist_test_all_stars[features]
+y_test = topic_dist_test_all_stars['Star'] 
+
+classifiers = [MultinomialNB(), LogisticRegression(), RandomForestClassifier(n_estimators=100, n_jobs=2), AdaBoostClassifier(n_estimators=100)]
+classifiers_names = ['Multinomial Naive Bayes', 'Logistic Regression', 'Random Forest', 'AdaBoost']
+
+LdaSentimentResults = {}
+for (i, clf_) in enumerate(classifiers):
+    clf = clf_.fit(x_train, y_train)
+    preds = clf.predict(x_test)
+    
+    precision = metrics.precision_score(y_test, preds)
+    recall = metrics.recall_score(y_test, preds)
+    f1 = metrics.f1_score(y_test, preds)
+    accuracy = accuracy_score(y_test, preds)
+    report = classification_report(y_test, preds)
+    matrix = metrics.confusion_matrix(y_test, preds, labels=starsGroup.groups.keys())
+    
+    data = {'precision':precision,
+            'recall':recall,
+            'f1_score':f1,
+            'accuracy':accuracy,
+            'clf_report':report,
+            'clf_matrix':matrix,
+            'y_predicted':preds}
+    
+    LdaSentimentResults[classifiers_names[i]] = data
+
+
+
+cols = ['precision', 'recall', 'f1_score', 'accuracy']
+pd.DataFrame(LdaSentimentResults).T[cols].T
+
+for model, val in LdaSentimentResults.iteritems():
+    print '-------'+'-'*len(model)
+    print 'MODEL:', model
+    print '-------'+'-'*len(model)
+    print 'The precision for this classifier is ' + str(val['precision'])
+    print 'The recall for this classifier is    ' + str(val['recall'])
+    print 'The f1 for this classifier is        ' + str(val['f1_score'])
+    print 'The accuracy for this classifier is  ' + str(val['accuracy'])
+    print 'The confusion matrix for this classifier is  \n' + str(val['clf_matrix'])
+    print '\nHere is the classification report:'
+    print val['clf_report']

rating_prediction_tfidf.py

@@ -0,0 +1,60 @@
+from sklearn import metrics
+from sklearn.metrics import classification_report
+from sklearn.naive_bayes import MultinomialNB
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.linear_model import LogisticRegression
+from sklearn.metrics import accuracy_score
+
+from rating_prediction_lda import starsGroup,all_text_train, all_text_test,topic_dist_train_all_stars,topic_dist_test_all_stars
+
+# Extracting features using term frequency
+tfidfvectorizer = TfidfVectorizer()
+
+tfidfTrain = tfidfvectorizer.fit_transform(all_text_train)
+tfidfTest = tfidfvectorizer.transform(all_text_test)
+
+tfidfLabelTrain = topic_dist_train_all_stars['Star']
+tfidfLabelTest = topic_dist_train_all_stars['Star']
+
+classifiers = [MultinomialNB(), LogisticRegression()]
+classifiers_names = ['Multinomial Naive Bayes', 'Logistic Regression']
+
+TFIDF_Pred_Results = {}
+for (i, clf_) in enumerate(classifiers):
+    clf = clf_.fit(tfidfTrain, tfidfLabelTrain)
+    prediction = clf.predict(tfidfTest)
+    
+    precision = metrics.precision_score(tfidfLabelTrain, prediction)
+    recall = metrics.recall_score(tfidfLabelTest, prediction)
+    f1 = metrics.f1_score(tfidfLabelTest, prediction)
+    accuracy = accuracy_score(tfidfLabelTest, prediction)
+    report = classification_report(tfidfLabelTest, prediction)
+    matrix = metrics.confusion_matrix(tfidfLabelTest, prediction, labels=starsGroup.groups.keys())
+    
+    data = {'precision':precision,
+            'recall':recall,
+            'f1_score':f1,
+            'accuracy':accuracy,
+            'clf_report':report,
+            'clf_matrix':matrix,
+            'y_predicted':preds}
+    
+    TFIDF_Pred_Results[classifiers_names[i]] = data
+
+cols = ['precision', 'recall', 'f1_score', 'accuracy']
+pd.DataFrame(TFIDF_Pred_Results).T[cols].T
+
+
+
+for model, val in TFIDF_Pred_Results.iteritems():
+    print '-------'+'-'*len(model)
+    print 'MODEL:', model
+    print '-------'+'-'*len(model)
+    print 'The precision for this classifier is ' + str(val['precision'])
+    print 'The recall for this classifier is    ' + str(val['recall'])
+    print 'The f1 for this classifier is        ' + str(val['f1_score'])
+    print 'The accuracy for this classifier is  ' + str(val['accuracy'])
+    print 'The confusion matrix for this classifier is  \n' + str(val['clf_matrix'])
+    print '\nHere is the classification report:'
+    print val['clf_report']
+