BLR: Bounded Logistic Regression for Inspecting the Opinions on Indian Government Schemes

P.K. Kumar and Dr.S. Nandagopalan

Social media such as twitter, linked-in, blogs, facebook and so on, have become useful platform for the people to express their perspective of opinions on the development of the society. Analysing these opinions has gained more research interest due its importance in understand the people and take necessary decision for development. To analyze the opinions of the sentiment analysis is the widely used technique, which applies Natural Language Processing (NLP), Machine Leaning (ML) to understand the input text in terms of positive, negative and neutral opinions. It is highly complex to analyse the input text expressed by the user in social media due to its uncertainly, incompleteness nature of the context. In this paper an novel bounded logistic regression is proposed and investigated with Random Forest (RF), Decision Tree (DT) and Support Vector Machine (SVM) approaches with different Indian government schemes twitter dataset like Goods and Services Tax (GST), Demonetarization and Clean India. From the obtained results, proposed approach gives the better prediction accuracy compared to existing techniques.

Volume 11 | Issue 1

Pages: 404-413