Conducting Sentiment Analysis on Twitter Tweets to Predict the Outcomes of the Upcoming Karnataka State Elections

Abstract

- This research paper aims to predict how Twitter tweets from a specific politician correlate to their winning a seat in a state election. To understand this effect, sentiment analysis has been conducted on tweets by politicians in Karnataka to help predict who will win in the upcoming 2023 Karnataka Legislative Assembly election. Though previous research has already been done in this area, most studies have only focussed on the sentiment analysis of tweets. This paper goes further as it also looks at other factors, including the number of retweets and comments a tweet garners, which measures the tweet's engagement. A model has been created that weighs each factor to help predict who will win an election for a particular constituency. Through this model, a 72.7% accuracy has been achieved. However, the Twitter API severely limited the quantity and quality of data collected. These results can be expanded to help predict elections for other states. They could potentially help understand the effect of positive and negative sentiment on the winnability of a political candidate.