Two-Class Classification: Comparative Experiments for Chronic Kidney Disease

Abstract

Over two million of population across worldwide is currently depending on dialysis treatment or a kidney transplant to survive from kidney disease. Therefore, it is imperative for health agencies such as hospitals or insurance companies to predict the probabilities of patients who suffers from chronic case of kidney diseases, hence requiring medical attentions. This study performs a comparative experiment on prediction of chronic kidney disease via a classification methodology. Two supervised classification algorithms are used to build the classification model, which are Two-Class Decision Forest and Two-Class Neural Networks. Experimental results showed that Neural Network performed better based on all features but Decision Forest produced optimal performance with high accuracy, and precision as compared to Neural Networks and other algorithms from the literature such as K-Nearest Neighbor, Support Vector Machine, and Rule Induction.