Skip to content

Latest commit

 

History

History
12 lines (12 loc) · 705 Bytes

File metadata and controls

12 lines (12 loc) · 705 Bytes

SentimentAnalysis-Rotten-tomatoes

Sentiment Analysis on Rotten Tomatoes dataset of 150,000 reviews .

Refer FinalMovieReview notebook

1->Tokenized the reviews

2->Removed punctuations stopwords and duplicate reviews

3->Used CountVectorizer and Tf-Idf transformation to convert reviews to word-vectors

4->Created Pipeline for predictions

5->Used vader for sentiment analysis which didn't performed well

6->Used LinearSVC (accuracy 0.81)

7->Used LogisticRegression (accuracy 0.80)

8->Used RandomForestClassifier (accuracy 0.80)

9-> Used LSTM for predicting exact rating between 1 to 5 but got accuracy of 0.43