<![CDATA[Social media has become a primary platform for expressing opinions on the performance of public officials, including Nadiem Makarim, the Minister of Education, Culture, Research, and Technology. Opinions on Twitter reflect diverse public perceptions, making sentiment analysis essential to understanding these trends. This study aims to analyze public sentiment toward Nadiem Makarim’s performance and optimize sentiment classification models in handling data imbalance. The methodology employs a Support Vector Machine (SVM) with Term Frequency-Inverse Document Frequency (TF-IDF) through three scenarios: tuning TF-IDF parameters, selecting the best SVM kernel, and applying the Synthetic Minority Oversampling Technique (SMOTE) to address data imbalance. Experimental results indicate that the combination of max_features = 2000 and min_df = 2 yields the best F1-score of 68%, with the linear kernel being the most stable. Although SMOTE successfully balances class distribution, accuracy slightly decreases from 68% to 66%.]]>
Copyrights © 2025
