Klasifikasi Sentimen Terhadap Kebijakan Tapera Menggunakan Komparasi Machine Learning dan SMOTE

Abstract

The Indonesian government's Public Housing Savings Program (Tapera) aims to help low- and middle-income persons get housing financing. Although the initiative strives to satisfy housing requirements, the public has responded in a variety of ways, as evidenced by social media posts. The goal of this study is to examine public sentiment towards the Tapera policy using YouTube comment data to provide an overview of popular perspective. This study combines sentiment analysis with machine learning algorithms, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Multinomial Naïve Bayes (NB), and Decision Tree. The data is divided into three scenarios, namely 70% training data and 30% test data, 80% training data and 20% test data, and 90% training data and 10% test data. Data balancing is also performed with SMOTE.  The performance evaluation is based on each algorithm's accuracy, precision, recall, and F1 Score values. The results showed that the SVM algorithm performed the best in all circumstances, with the greatest accuracy of 88% and an F1 score of 85%. The multinomial Naïve Bayes algorithm ranked second with steady accuracy, whereas KNN and Decision Tree had poorer performance. This suggests that SVM is the most effective method for processing sentiment data regarding Tapera policy.