Since the appearance of the World Wide Web, there has been a significant change in the way we share information. People can express themselves on the web through many mediums. Examples of such mediums are Blogs, wikis, forums and social networks, where in these platforms being mentioned, people are able to exercise posting and sharing of opinions, thereby leading to the need for research in sentiment analysis. This paper contributes to the field of Sentiment Analysis of electoral prediction using Twitter, which aims to extract opinions from text using machine learning classifiers such as Naive Bayes, Max Entropy and SVM. The problem statement for this research is the classification of textual information as whether it bears positive sentiment or negative sentiment. Our research focuses on a novel data pre-processing method which enhances the performance of different supervised algorithms. With our method, the accuracy of Naive Bayes was enhanced to 89.78% (hybrid), Maximum Entropy was enhanced to 87% (hybrid), SVM was enhanced to 91% (unigram), and ensembling of Naive Bayes and SVM was enhanced to 95.3% (bigram).