Sentiment Analysis have been done on twitter data regarding stock market using Naive Bayes Classifier. We have tested a few feature selection techniques to improve the accuracy of Naive Bayes Classifier. The feature selection techniques tested are: TF-IDF, Word Frequency, Document Frequency, Sparsity Reduction and Chi Square Statistics. The code has been implemented in R. The dataset that has been used for implementation is provided in stock_stat.csv. The source of dataset: https://www.kaggle.com/yash612/stockmarket-sentiment-dataset