Analysis of Chronic Kidney Disease Prediction Using Decision Tree and K-Nearest Neighbor Classification
##plugins.themes.bootstrap3.article.main##
Abstract
Chronic Kidney Disease (CKD) is a major threat in medical analysis and is one of the major contributors of death as a non-communicable disease, affecting 10 to 15 percent of the worldwide population. Accurate detection of CKD in its initial stages is thought to be critical for minimizing the effects of health complications of the patients such as hypertension, iron deficiency, bone disorders occurring due to imbalance of minerals, malnourishment, pH fluctuations and abnormalities, and neurological complications through timely intervention with appropriate treatments. This study offers the steps for predicting status of CKD using medical records which comprise the data preprocessing, managing missing values, and feature extraction. Several studies on the detection of CKD at an early stage have been conducted using machine learning techniques. The two classification models such as Decision Tree (DT) and K-Nearest Neighbor (KNN) is used in this study. The performance of each classification methods is compared with each other and identified that KNN is the best classifier. Also using the relevant data extracted from the clinical dataset after feature extraction, the patient is said to have CKD or not CKD based on KNN approach with accuracy of 97 percent.