Analyzing Public Sentiment Toward Prabowo Subianto on X: A Multinomial Naive Bayes Approach with SMOTE

Abstract

Social media platform X serves as a major source of public opinion on political figures, including Indonesian President Prabowo Subianto. This study aims to conduct sentiment analysis on X user posts related to Prabowo Subianto during the period from June 2023 to June 2024 using the Multinomial Naive Bayes algorithm. The dataset consists of 42,199 posts collected through crawling X. The data were processed through text preprocessing stages, including cleansing, tokenization, normalization, stopword removal, stemming, and text translation, followed by feature weighting using Term Frequency–Inverse Document Frequency (TF-IDF). This study compares model performance with and without data balancing using the Synthetic Minority Over-sampling Technique (SMOTE) and applies K-Fold Cross-Validation for model validation. Model evaluation uses accuracy, precision, recall, and F1-score metrics. The results show that applying SMOTE improves model performance and enables more balanced sentiment classification. Overall, the Multinomial Naive Bayes algorithm proves effective for sentiment analysis of X social media data related to President Prabowo Subianto during the 2023–2024 period.