Comparative analysis of classification models in diagnosis of type 2 diabetes

摘要

Diabetes has become one of the major causes of premature diseases and death in most countries. A major problem in medical science and bioinformatics analysis is to obtain the correct diagnosis on specific important information. In general, several tests are done that includes clustering or classification on large scale of data. However, many tests might complicate the main diagnosis process and lead to difficulty in getting the final results. Machine learning techniques are used to build models to overcome this kind of difficulty. Therefore, there are two main purposes of this study. First, implementing classification models to diagnose diabetes efficiently and easily. Second, investigating and comparing the performance of different classification models. In this study, we proposed Fuzzy Expert System (FES) that used Fuzzy Inference System (FIS) model for incidence of diabetes. We applied two common classification algorithms which are logistic regression and support vector machine on Pima Indian Dataset to compare them with our proposed FIS model. In order to perform our experiment, we used two data mining tools named WEKA and MATLAB.