Twitter Sentiment Analysis Using Machine Learning

Abstract

Sentiment analysis is the process of identifying and categorising the emotions expressed in text. When tweets are analysed, they typically generate a large amount of sentiment data. This information allows us to better understand people's perspectives on a variety of issues. This study tries to classify tweets based on their sentiment. They can express either positive or negative emotions. Twitter is a social networking and micro blogging platform that allows users to post 140-character status updates or opinions. It has about 200 million registered users, 100 million active users, and half of them log in every day, resulting in nearly 250 million tweets per day. Because of the widespread use, we want to reflect the prevalent attitude by analysing tweets. Predicting political elections and macroeconomic phenomena like stock exchanges necessitates a look at public sentiment. We attempt to categorise the tweets as positive or negative. To represent the "tweet," it must also extract valuable elements from the text, such as unigrams and bigrams. We use machine learning methods to analyse sentiment using the collected features. Individual models did not provide high accuracy on their own. So we created an Ensemble Model that predicts based on a majority vote using Naive Bayes, Logistic Regression, and Support Vector Machines.