Sentiment Analysis on Social Media Using Machine Learning

The rapid growth of user-generated content on social media platforms has created a pressing need 
for automated tools to extract and interpret public opinion at scale. This paper presents a comparative 
evaluation of machine learning approaches for sentiment classification of social media posts. We experiment 
with Naïve Bayes, Logistic Regression, Support Vector Machine (SVM), Random Forest, and a bidirectional 
LSTM on a balanced sample of 50,000 Twitter posts drawn from the Sentiment140 dataset. Our results 
indicate that LSTM achieves the highest accuracy (88.3%) while linear SVM (85.6%) provides a compelling 
balance between performance and computational efficiency. We discuss the implications of these findings 
for real-world sentiment monitoring applications.