twitter_sentiment

This project aims at derving sentiments from the text in the tweet. The pre processing step removes all numerals along with handle names, basically all words that cannot be linked to any emotion. In this project, we compared two diffrent prediction algorithms, one Naive Bayes, and the other Text Blob. We iterated through the data and chose the 100 most frequent words as our features. We got 62% accuracy with Text Blob and around 50% accuracy with Naive Bayes.