Optimization the Naive Bayes Classifier Method to diagnose diabetes Mellitus

Abstract

World Health Organization (WHO) states that Diabetes Mellitus is the world's top deadly  disease. several studies in the health sector including diabetes mellitus have been carried out to detect diseases early. In this study  optimization of naive bayes classifier using particle swarm optimization was applied to the data of patients with 2 classes namely positive diabetes mellitus and negative diabetes  mellitus and data on patients with 3 classes, those who tested positive for diabetes mellitus type 1, diabetes mellitus type 2 and negative diabetes mellitus. After testing, the algorithm of Naive Bayes Classifier and Naive Bayes Classifier based on Particle Swarm Optimization, the results obtained are the Naive Bayes Classifier method for 2 classes and 3 classes each producing an accuracy value of 78.88% and 68.50%. but after adding Particle Swarm Optimization the value of accuracy increased respectively to 82.58% and 71, 29%. The classification results for 2 classes have an accuracy value higher than 3 classes with a difference of 11.29%