Author(s): Dr. J. Jegathesh Amalraj, Dr. M. Sivakumar
Diabetes Mellitus has emerged as a worldwide epidemic. A number of people will appreciate if diabetic diagnostic aid is provided by using a set of data with only medical information and not with any advanced medical equipment. This can make a huge positive impact in the life of lot of people. The aim of this study is to test the diabetes data in two cases as with and without pre-processing and observe the difference in the classification accuracy. Depending upon the results obtained, further decision is made. It is observed that there is increase in accuracy in classification after pre-processing since unwanted data always lead to a decrease in the performance or classification accuracy.