An Approach for Diabetes Detection using Data Mining Classification Techniques

Disease diagnose by expert systems, is one of the areas where tools of data mining are establishing successful results. The aim of this paper is to discover solutions for diagnosing the disease by analyzing the patterns found in the data through techniques of data mining like classification analysis. Classification is a common technique used in data mining that utilizes a set of pre-classified examples for developing a model that can help in classifying the population of records at enormous amount. There are various techniques of classification that are used for analysis of biomedical data. These include Naive Bayes, Bayes Net, J48, SMO, and Random Forest. In this paper, the comparison of different classification algorithms using Weka has been shown. Also these techniques are used to find out which algorithm is most suitable. The best algorithm based on the Cross validation is SMO classifier with an accuracy of 77.34 % and has the lowest average error at 22.65 % compared to others. The best algorithm based on the Percentage split, Decision Table classifier with accuracy of 81.99 % and has the lowest average error at 18.00 % compared to others.