CHRONIC KIDNEY DISEASE DETECTION USING ENSEMBLE LEARNING TECHNIQUES AND COMPARATIVE STUDY

Abstract

A common health problem around the globe, chronic kidney disease (CKD) must be identified early in order to be effectively managed. The accuracy of CKD diagnosis may be increased with the use of machine learning approaches, especially ensemble learning. In order to determine which model performs best for CKD detection, this research will compare and contrast several ensemble learning strategies. Ten distinct models are evaluated in the study: Bagging, Random Forest, Gradient Boosting, Ada Boosting, XGBoost, K-Nearest Neighbours (KNN), Decision Tree, Decision Tree after Pruning, Logistic Regression, and Linear Discriminant Analysis. A CKD dataset is used to evaluate these models based on criteria including accuracy, precision score, and recall score. The comparative study results demonstrate how ensemble learning techniques might raise CKD detection accuracy. The findings provide crucial details about the optimal model for CKD detection, which can help with early diagnosis and better patient outcomes.