COMPARISON OF THE PERFORMANCE OF NAÏVE BAYES AND CORRELATED NAÏVE BAYES METHODS WITH THE APPLICATION OF SYNTHETIC MINORITY OVER-SAMPLING TECHNIQUE

Abstract

Classification is the process of creating a model to recognize patterns with the aim of mapping them into specific classes and predicting classes. Naive Bayes is a popular, simple and effective classification method with a probabilistic approach based on Bayes' Theorem. The assumption of independence in this method sometimes makes the classification performance decrease. Correlated naïve bayes corrects this assumption by considering attribute correlations, while SMOTE is used to overcome data imbalances. This approach is important in medical data analysis, one of which is predicting ischemic heart disease. This study aims to compare the performance of Naïve Bayes and Correlated Naïve Bayes methods in the classification of ischemic heart disease, with the application of SMOTE to overcome data imbalance. The analysis was carried out using ischemic heart disease data at the Integrated Heart Center of Dr. Wahidin Sudirohusodo Hospital, Makassar City, for the period of July 2021 to July 2022. Naïve Bayes managed to classify 66 data with 75% accuracy, 94% precision, and 62% sensitivity. Meanwhile, Correlated Naïve Bayes showed better performance by correctly classifying 77 data, resulting in 87.5% accuracy, 86% precision, and 94% sensitivity. These results show that Correlated Naïve Bayes has a superior performance in classifying ischemic heart disease.