IMPROVING THE PERFORMANCE OF SUPPORT VECTOR MACHINE WITH FORWARD SELECTION FOR PREDICTION OF CHRONIC KIDNEY DISEASE

Abstract

Chronic kidney disease is a disorder that affects the kidneys and arises due to various factors. Chronic kidney disease, usually develops slowly and is chronic. For prevention and control, proper treatment is needed, so that detection of this disease can play a very important role. This study aims to determine the level of accuracy in predicting chronic kidney disease through SVM based on forward selection and to determine the performance of Feature Selection which is applied to the SVM method in solving problems in chronic kidney disease. This research was conducted an experiment on the SVM method using various kinds of kernels and it was seen that SVM with the dot kernel was 98.50% with AUC 1,000 which was superior to the polynominal kernel and RBF. However, when the experiment was carried out again by applying FS to SVM, it was found that SVM+FS with the RBF kernel outperformed the other kernels by 99.75% with AUC 1,000. So it can be concluded that the Forward Selection of SVM has succeeded in improving its performance, especially in this case, namely the prediction of chronic kidney disease