Sentiment analysis of user comments using TF-IDF and machine learning classifiers: A comparative study
Mrinmoy Kayal, Law Kumar Singh, Jayadeep Pati · 2025
Due to social media, text mining queries have exploded. Text analysis, especially online, is growing in popularity. This trend should continue. Social media networks generate a lot of text data, allowing users to freely comment. The rise of social media contributes to this. Many commercial applications require comment analysis. NLP Sentiment Analysis (SA) is essential for identifying feelings in reviews and comments. This is one of the most crucial steps. This study describes how to use the TF-IDF vectorizer to create a feature extraction machine learning model. We want the greatest precision with this model. Sentiment analysis usually predicts user comment emotions. This drives the strategy. This proposal&s;s principal goals fall into three areas. We classified a dataset as good or negative. Our process started here. The second step entailed exhaustively reviewing six classification systems. We ranked classifier performance using Decision Tree, Random Forest, Logistic Regression, Gaussian NB, AdaBoost, and XGBoost. We used a different order for each performance. We conclude with the Borderland Emotion Dataset to demonstrate our random forest model&s;s effectiveness and best outcomes.