Prediction of Kidney Disease using Hybrid System based on Regression, Decision Tree and K-Near Neighbour method

Abstract: In the world, many people are suffering from kidney failure, and a large percentage of people are not aware until the advanced stage. Therefore, early detection and diagnosis of disease can be an effective step in the treatment and survival of the patient. In this study, the proposed model is a hybrid system consisting of decision tree, linear regression and nearest neighbour. The first step is created based on the training dataset of the decision trees models, linear regression and nearest neighbour. In the second step, the related class of the available samples in the test dataset is predicted based on the obtained models. In the last step, final output is determined based on the voting strategy. The obtained accuracy of this method is 89.47% in the proposed method, which had the best performance in comparison with the decision tree, linear regression and nearest neighbour.

Keywords: Decision Tree, Regression, K-NN, Hybrid Methods, Kidney Disease.