This project performs a sentiment analysis on the amazon kindle reviews dataset using python libraries such as nltk, numpy, pandas, sklearn, and mlxtend using 3 classifiers namely: Naive Bayes, Random Forest, and Support Vector Machines. The categories to be classified into are positive, negative and neutral reviews. Finally the obtained outputs are compared with the expected output using the f1-score computation, for each classifier and the decision boundaries created by the SVM are plotted. The kindle reviews dataset can be found at the following website: http://jmcauley.ucsd.edu/data/amazon. Select and download the kindle store 5-core reviews dataset, under small subsets for experimentation. Details of the procedure can be referred here: https://journalofbigdata.springeropen.com/articles/10.1186/s40537-015-0015-2#Sec3
The code outputs of the first 50000 reviews are included in the uploaded files.