Analisis sentimen kebijakan Tabungan Perumahan Rakyat menggunakan model Support Vector Machine dan Multinomial Naive Bayes

Abstrak

TAPERA (Public Housing Savings) is a government program implemented on May 20, 2024, under Law No. 24 of 2020 and Government Regulation No. 21 of 2024. Since its introduction, this program has sparked public debate regarding its potential benefits, financial implications, and perceived disadvantages among citizens. This research analyzes public sentiment toward TAPERA by applying two machine learning algorithms: Support Vector Machine (SVM) and the Naïve Bayes algorithm. A total of 3,168 comments were collected from the YouTube platform and automatically labeled using the InSet Lexicon-based approach to classify sentiments into positive, neutral, and negative categories. The data were split into 80% for training and 20% for testing. Model performance was evaluated using a confusion matrix. The results indicate that SVM achieved 88.93% accuracy, outperforming Naïve Bayes, which achieved 65.39%. Furthermore, sentiment analysis reveals that negative sentiment dominates public opinion, followed by a substantial proportion of neutral sentiment. These findings highlight public concerns about TAPERA and demonstrate the effectiveness of SVM for sentiment classification.