Analysis of Public Sentiment Toward the Free Nutritious Meals Program Using the Naive Bayes Algorithm

Abstract

The increasing use of digital platforms as spaces for expressing public opinion has created opportunities for applying natural language processing techniques to evaluate societal responses toward government policies. This study aims to analyze public sentiment toward the Program Makan Bergizi Gratis (MBG) using the Multinomial Naive Bayes algorithm. The research applies an empirical machine learning approach involving data preprocessing, lexicon-based sentiment labeling, feature extraction using Count Vectorizer with unigram and bigram representation, and model evaluation through accuracy, precision, recall, F1-score, and confusion matrix analysis. The dataset consists of 3,300 public comments processed through a structured computational pipeline and divided using stratified sampling into training and testing data. The findings indicate that public sentiment is dominated by negative responses, followed by neutral and positive categories, reflecting critical public evaluation toward program implementation aspects. The classification model demonstrates reliable performance in identifying sentiment patterns and provides an analytical framework for understanding digital public perception. This study contributes to computational social analysis by integrating machine learning techniques with policy-oriented sentiment monitoring.