Election Forecasting with Machine Learning and Sentiment Analysis: Karnataka 2023

Data science is rapidly transforming the political sphere, enabling more informed and data- driven electoral processes. The ensemble machine model which is made up of Random Forest Classifier, Gradient Boosting Classifier, and Voting Classifier, introduced in this paper makes use of machine learning methods and sentiment analysis to correctly forecast the results of the Karnataka state elections in 2023. Election features such as winning party, runner- up party, district name, winning margin, and voting turnout are used to evaluate the effectiveness of different machine learning paradigms. Similarly, it also makes use of sentiment analysis through party tweet and public reactions for further breaking down reliance upon past elections data alone. This study demonstrates that using both past historical records and current public opinion yields precise predictions about how electable leaders are. This reduces reliance on a historical dataset. The experimented results shows that, how machine learning and sentiment analysis can predict election results and provide useful data for election decision making. We compared various machine learning models in this study, including logistic regression, Grid SearchCV, XGBoost, Gradient Boosting Classifier, and ensemble model. With an accuracy of 85%, we demonstrated that our ensemble model outperformed machine models such as XGBoost and Gradient Boosting Classifier. It also offers a novel method for predictive analysis.