Twitter is the most generally used micro blogging and social media platform for expressing ideas. Today's society makes it challenging to analyses Twitter sentiment. In the area of opinion mining research, the utilization of sentiment classification for the English language has increased in popularity. It is still based research on specific Indian languages, and progress is really slow. In this paper, we present a useful technique for evaluating the sentiment of Gujarati tweets on Twitter. This analysis is used to determine and classify Gujarati tweets into three classes (Positive, Negative and Neutral). We concentrate on extracting features and word identification because pre-processing methods improve classification accuracy.