Support Vector Machine with Purified K-Means Clusters for Chronic Kidney Disease Detection

Abstract

Chronic kidney disease is a kidney disease in which there is a function loss of kidney and it is occurred overtimes and years. This disease is perceptible until the kidney losses 25% of its function. Chronic kidney disease requires a correct and appropriate diagnostic process in order to provide relevant and proper treatment which is in accordance with the diagnosis. Using current developed technology, the diagnosis process can be easily conducted. The diagnosis can be carried out by employing data mining techniques such as clustering and classification. This study seeks to explore the implementation of the K-Means algorithm as a clustering algorithm and Support Vector Machine algorithm as a classification algorithm. Clustering process is used to determine data on the pure cluster then the data will be classified using the Support Vector Machine algorithm. In the classification process with the Support Vector Machine algorithm, various non-linear kernels such as polynomial kernels, RBF kernels, and sigmoid kernels are used. Based on the research results, the highest accuracy is obtained from the classification process with two clusters, which is 100% in all kernel functions. As for the highest accuracy in the classification with three clusters, four clusters, and five clusters are generated by the classification process using the RBF kernel.