<![CDATA[The Mens Rea stand-up comedy performance by Pandji Pragiwaksono has generated diverse public responses on the X social media platform, reflecting positive, negative, and neutral sentiments. This study aims to analyze public sentiment toward the performance using the Multinomial Naive Bayes algorithm. Research data were collected through a crawling process on X, resulting in 3,019 tweets. The research stages included data selection, sentiment labeling using a lexicon-based approach validated by an expert, preprocessing (cleansing, case folding, tokenizing, stopword removal, and stemming), TF-IDF term weighting, and sentiment classification using Multinomial Naive Bayes. Model performance was evaluated using a Confusion Matrix with five train-test split scenarios (90:10, 80:20, 70:30, 60:40, and 50:50). The best performance was achieved using the 80:20 split, with an accuracy of 85.79%, while the highest F1-score of 0.9231 was obtained for the neutral sentiment class. Overall, the combination of TF-IDF and Multinomial Naive Bayes demonstrated stable and effective performance for public sentiment classification of the Mens Rea stand-up comedy on the X platform.]]>
Copyrights © 2026
