Performance Comparison of Naive Bayes and Support Vector Machine (SVM) Methods in Public Sentiment Analysis of the Free Nutritious Meal Program (MBG)

Perbandingan Kinerja Metode Naive Bayes dan Support Vector Machine (SVM) dalam Analisis Sentimen Publik terhadap Program Makan Bergizi Gratis (MBG)

Abstract

The Free Nutritious Meal Program (MBG) launched by the Indonesian government has triggered diverse public discourse on social media. This study compares Naive Bayes and Support Vector Machine (SVM) in classifying public sentiment toward the MBG program using 1,050 valid Indonesian-language tweets collected via web scraping during January–December 2025. Text preprocessing included case folding, cleaning, tokenizing, stopword removal, and stemming using Sastrawi, followed by TF-IDF feature extraction, yielding 2,920 features. The dataset was split into 840 training and 210 testing samples. Results show Naive Bayes achieved slightly higher scores than SVM across all metrics, with 70.48% accuracy, 71.14% precision, 70.48% recall, and 69.73% F1-score, versus SVM's 68.57% accuracy. Naive Bayes was also far more computationally efficient, training about 52 times faster than SVM. Sentiment distribution was relatively balanced (Neutral 34.2%, Positive 33.0%, Negative 32.8%), indicating the public remained in a critical evaluation phase toward the program.