The Sentiment Analysis of Public Comments on “Program Makan Siang Gratis” Using KNN (K-Nearest Neighbor) and SMOTE Algorithm

Abstract

This research analyzes public sentiment on “Program Makan Siang Gratis” using the KNN algorithm, enhanced with the SMOTE technique, to provide insights and recommendations for policymakers aiming to achieve “Indonesia Emas 2045”. The study employs Google Colab and Python for testing. KNN is used for sentiment analysis and classification, with SMOTE addressing data imbalance. Results from two scenarios without SMOTE and with SMOTE show that performance is more optimal without SMOTE, as SMOTE decreases performance by 34%. The k4 parameter yields the best results: 76% accuracy, 57% precision, 77% recall, and 65% F1-Score. Analysis of comments from the “tempodotco” YouTube channel reveals that public sentiment towards the program proposed by President Prabowo Subianto and Vice President Gibran Rakabuming Raka is predominantly negative