Diabetes is an affecting people disease nowadays. Over 246 million people worldwide with a majority of them being women had been affected by diabetes. According to the World Health Organization (WHO), by 2025, this number is expected to increase to over 380 million. The disease has been ranked as the fifth deadliest disease in United States with no imminent cure in sight. Along with the increasing of the information technology and its continued advent into the medical and healthcare sector, the cases of this disease and the symptoms are well documented. This paper aims at finding the best performance by four classification algorithms, which are Naive Bayes, Simple Logistics, REPTree, and Sequential Minimal Optimization (SMO). Model testing on 200 tuples with 9 attributes in the diabetes dataset revealed that Naive Bayes achieved highest accuracy of 85%. To improve the overall accuracy, it is necessary to use more data set with larger number of attributes and use a better feature selection method in future works.
