Anima Srivastava & Tanveer J. Siddiqui

Department of Electronics and Communication, University of Allahabad.

Abstract

The initiation of Goods and Service Tax was one of India's most notable economic revolutions leading to extensive debates. It ushered a massive platform via Social Media websites for the general public to throw spotlight on their opinions of GST. Ascertaining public sentiment with these opinions will help harness future reforms. We accumulated GST-related tweets from last day of April 2018 until 1st of May 2018, and analyzed their polarization stimulated with this detailing. We also studied six different classifiers: Ridge Classifier, Linear SVC, Logistic Regression, Perceptron, K-Nearest Neighbor and Decision Tree. Using tf-idf feature form test runs were conducted for each classifier. The effectiveness of each scenario was assessed and found that ridge classifier observed the maximum accuracy 96%. The performance assessment is made with other available work on GST. All classifier performed better with tf-idf feature with the existing feature permutations also findings are compared with other existing works on identical dataset.

Keywords- Goods and Service Tax (GST), linear classification, Supervised Machine Learning.