Amazon review data is available here : http://jmcauley.ucsd.edu/data/amazon/
After you get the review dataset:
- Run the reading_review_data.py file to read the GZ fil. This will create a csv file containing the review data.
- Next run the Features_Extraction.py file to extract text features such as TF_IDF and Latent semantic analysis features which will be given as an input to our classifier.
- Finally run the Classifier.py which takes the Text_Features.csv generated in step-2 to perform the classification using Random Forest and Gradient Boosting classifers.
Note : Extreme gradient boosting classifier code is also included, using the XG_Boost_Classifier.r you can use xg_boost classifier.
My model acheived an accuracy of 78.59% in classifying the helpful and unhelpful reviews using the Amazon Dataset.