Comparative analysis of classification models for diagnosis Type 2 Diabetes

Abstract

Diabetes mellitus type 2 has become one of the major causes of premature diseases and death in many countries. It accounts for the majority of diabetes cases around the world. Thus, we need to develop a system that diagnoses type 2 diabetes. In this thesis, a fuzzy expert system is proposed using the Mamdani fuzzy inference system to diagnose type 2 diabetes effectively. In order to evaluate the performance of our system, a comparative study has been initiated, and will contrast the proposed system with data mining algorithms, namely J48 Decision tree, multilayer perceptron, support vector machine, and Naïve Bayes. The developed fuzzy expert system and the data mining algorithms are validated with real data from the UCI machine learning datasets. Moreover, the performance of the fuzzy expert system is evaluated by comparing it to related work that used the Mamdani inference system to diagnose the incidence of type 2 diabetes. Alternate title: Comparative analysis of data mining algorithms for diagnosis Type 2 Diabetes