This thesis project addresses the problem of sentiment analysis of political speeches given by the leaders and we also find out the impact of speeches on elections. After last general election in India, the political landscape of the India has undergone through a significant changes which are speculated to stem mass adoption of social media. The political parties and there top leader are using social media network as a primary way to connect with the peoples. In this modern era of communication each and every one connected to the internet. And they tend to express their stands on social media website about products, movies, sports, social matter and even on government policies. With more and more people coming online, social media becoming one of the top medium for opinion sharing. Many experts believed that social media is playing a crucial role on elections. In current scenario a huge population in India is expressing their view and opinion for government policies or for any social issue through social media. Politicians are frequently accused of changing their stance on issues or going along with the party’s stance. We wanted to see whether politician’s speech differed greatly depending on their party affiliation, or when or where they were speaking. Are these speeches have any significant effects on election results or not. In this thesis we study sentiment analysis on speech, tweets and the reaction of public as well, by extracting tweets from twitter with the help of twitter API, and speeches which are available in public domain. In this project I used machine learning algorithm naïve bayes and support vector machine algorithms for classification and NLP (Natural Language processing).