A Fine-tuned Gradient Boosting Method to Improve Sentiment Classification of Lockdown Extension Tweets
Jayalakshmi P. V, M. Lakshmi · 2022 International Conference on Electronics and Renewable Systems (ICEARS) · 2022
Social media is one of the sources that handles huge amount of data on an extraordinary scale. People sharing their own ideas, thoughts, views related to the current topic or trending topic using the biggest platform in social media like Twitter, Facebook, etc. As unforeseen as the event of Covid infection 2019 (COVID-19) was, it has been fundamentally influencing individuals everywhere on the world, there is a demand to examine and analyze the individuals on the endemic COVID-19. This paper concentrates on the sentiment analysis of COVID-19 data which was extracted from twitter using python language and analyzed by machine learning algorithms and to predict people’s reaction towards lockdown extension, what the careful steps they have to take are and whether individuals are following government's rules and so on. Tweets have been collected from Twitter using Tweepy API. After extracted, the text was classified with the help of Vader, and it was trained and tested with 6 machine learning algorithms to find the accuracy and prediction about the public’s view. After the comparison of those 6 algorithms and found the best and fine-tuned some features of Gradient Boosting algorithm to enhance the prediction. This study concludes that majority of Indian peoples supported the Government for taking such a decision to take care of themselves.