Perbandingan Metode Naïve Bayes dan Support Vector Machine dalam Analisis Sentimen Citra Polri berdasarkan Opini pada Platform Twitter/X

Abstract

The Indonesian National Police (Polri) is a law enforcement agency responsible for maintaining security and order in Indonesia. In the digital era, Polri’s image has increasingly been highlighted on social media platforms such as Twitter/X, which serve as a major channel for the public to express opinions and criticism. This study aims to compare the performance of the Naive Bayes method and Support Vector Machine (SVM) in sentiment analysis of public opinion toward Polri. Naive Bayes, known for its probabilistic approach, is compared with SVM, a robust machine learning algorithm capable of classifying data with clear margins between classes. The dataset was divided into 80% training data and 20% testing data with stratification to ensure balanced sentiment proportions. Performance evaluation was conducted using accuracy, precision, recall, and F1-score through a confusion matrix. Results show that SVM achieved the highest accuracy of 90%, while Naive Bayes obtained 83%. In terms of F1-score, SVM reached a macro average of 0.90 with its best performance in the positive category (0.97), while Naive Bayes reached 0.83 with its best in the positive category (0.90). Overall, SVM outperformed Naive Bayes, particularly in classifying neutral sentiment. This study provides insights into the effectiveness of SVM for analyzing informal tweets and can serve as a reference for future research and public opinion monitoring system development.