Predicting the June 2019 Istanbul Mayoral Election with Twitter

In this study, it is aimed to predict the Istanbul Mayoral election which was held on 23 June 2019 with the data collected through the APIs provided by Twitter. Machine learning models were built on a sample data which was labeled manually. Then, these ML models were applied on the whole tweet data set to predict sentiments of tweets regarding candidates. Among applied ML methods, Logistic Regression provided best result for this study. Using predicted sentiments, voting rates for candidates are estimated following different approaches. 2.2% MAE for the voting rates was achieved with the best prediction.