What type of study is this?

This is a Quantitative Study study.

November 4, 2025Open Access

Performance Comparison of 10 Machine Learning Algorithms in Sentiment Classification on Platform X Regarding the Government’s Priority Program: Makanan Bergizi Gratis (MBG)

Key Points

Bagging achieved the highest accuracy at 81% in the sentiment classification task, indicating robust performance among the tested algorithms.
The evaluation was conducted using models like Naive Bayes and Gradient Boosting, with metrics such as precision and recall guiding the assessment.
Data preprocessing included tokenization and TF-IDF vectorization, ensuring the algorithms were optimized for sentiment analysis.
Findings may inform strategies for improving government communication and enhancing public perception of food programs.

Abstract

This study evaluates public sentiment on Platform X regarding the Government’s Priority Free Nutritious Food Program. A total of 2031 user comments were analyzed using 10 machine learning algorithms: Naive Bayes, Gradient Boosting, AdaBoost, Random Forest, Extra Trees, Logistic Regression, Linear SVM, SGD Classifier, Ridge Classifier, and Bagging. The dataset underwent preprocessing including lowercasing, stopword removal, stemming, and tokenization, followed by TF-IDF vectorization with 5000 features. Models were evaluated using accuracy, precision, recall, weighted F1-score, and 5-fold cross-validation. Bagging achieved the highest accuracy (81%) and weighted F1-score (81%), followed by Gradient Boosting (81%) and Random Forest (77%). Feature analysis revealed negative sentiment indicators such as ‘racun’, ‘stop’, ‘korupsi’, and positive indicators like ‘sehat’, ‘enak’, ‘bergizi’. These findings provide actionable insights for policy communication and program improvement.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper