<![CDATA[Sentiment analysis was conducted to examine public perceptions of the Free Nutritious Meal (Makan Bergizi Gratis, MBG) Program and identify prevailing opinion trends surrounding the policy. The analysis was initially conducted on 5,118 social media posts collected from X (formerly Twitter). After the preprocessing stage, which involved removing duplicate records, missing values, and irrelevant textual elements, a total of 4,332 posts remained for analysis. The resulting dataset was inherently subjective, as it comprised individual opinions reflecting diverse perspectives and styles of expression. Sentiment labeling was subsequently performed using a hybrid approach that combined lexicon-based labeling with manual annotation. The dataset comprised 50.42% positive, 38.55% negative, and 11.03% neutral sentiments. Among various text classification techniques, Support Vector Machine (SVM) was employed to classify sentiment in social media posts. The proposed framework comprised several sequential stages, including data collection, text preprocessing, hybrid sentiment labeling, word cloud visualization, feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF), model development, and performance evaluation. The classifier was trained and evaluated using three train-test split ratios (80:20, 70:30, and 60:40), followed by stratified 10-fold cross-validation to obtain a more robust performance assessment. The cross-validation results showed that all three data partitioning strategies achieved comparable performance. The 70:30 split produced the highest mean accuracy of 80.08 ± 1.38%, together with a weighted precision of 77.39%, a weighted recall of 80.08%, and a weighted F1-score of 76.19%.]]>
