Inferring political preference from Twitter tweets

Commercial popularity of social media and free availability of vast data has enhanced the interests of the researchers in analyzing its contents. Many businesses make use of such data to harness the mindset and likes of their target audience, thereby improving their profits. Sentiment analysis of Twitter texts have proved to be an effective way of voicing the needs of large masses and is used by many prominent politicians in making better campaigning strategies. Multiple machine learning classifiers are implemented in this study to access the stance of US citizens towards Democrats/Republicans to deduce which political party a user prefers from his tweets. Performance of a stacked ensemble is compared against a deep neural network for the mentioned problem domain.