A Comparative Analysis of the Performance of Naive Bayes, SVM, and Random Forest in Classifying SAR Information Communication Responses

Abstract

The development of information technology increases the need for fast, clear, and structured management of SAR information communication. Responses to SAR information may include appreciation, requests for information, and criticism; therefore, a text-based approach is needed to classify these responses more systematically. This study aims to analyze and compare the performance of Naive Bayes, Support Vector Machine (SVM), and Random Forest algorithms in classifying SAR information communication responses. The dataset used in this study consists of 500 synthetic records created based on SAR information communication response scenarios and does not represent actual public responses. The classification labels consist of three classes: appreciation, information request, and criticism. The research stages include synthetic data creation, text preprocessing, TF-IDF weighting, data splitting with an 80:20 ratio, model training, and model evaluation. The evaluation uses accuracy, precision, recall, F1-score, and confusion matrix. The results show that SVM and Random Forest achieve the highest accuracy of 95.00%, while Naive Bayes achieves 89.00%. Based on precision and F1-score, SVM provides the best performance with 95.66% precision and 95.04% F1-score. These findings indicate that machine learning algorithms can be used as an initial approach for classifying SAR information communication responses using synthetic data.