This document discusses user behavior analysis on social media data using sentiment analysis. It describes extracting data from social media platforms like Twitter using hashtags and keywords. The data is then preprocessed by removing URLs, symbols, stop words etc. Features like sentiment, unigrams, emoticons are extracted and classification algorithms like Naive Bayes, K-NN, Random Forest are used to classify the data based on sentiment. The results can help understand user behavior and opinions on various topics from their social media posts and comments.