Performance evaluation of Adtree, Functional Tree and lMT Classifiers with CFS Subset evaluator for Intelligent heart disease Prediction Devasena C lakshmi Department of IT and Operations, IBS Hyderabad, IFHE University, Hyderabad Online published on 5 May, 2018. Abstract Healthcare analytics an emerging research area, which analyses existing clinical data, provide better support for clinical practices and decision making. In healthcare analytics, medical data mining an important research field which provide support for healthcare practitioners and working as a third umpire for senior specialist, who gets confused in complex medical cases. Medical data mining (MDM) used to predict numerous diseases in which Heart disease is the chief cause of increasing mortality rate worldwide and is estimated to be the crucial cause of death globally by 2030[1]. Predicting the heart disease in early stage with some demographic and clinical data before going for Angiography and direct high level treatment will reduce the expense of the patient. However, the diagnosis of heart disease and its severity might be difficult for practicing medical doctors, who are not specialized in cardiovascular diseases. There are many papers in literature which can compare the existing algorithms in heart disease prediction[2–36], but none of the papers discussed tree based classifiers with correlation based feature subset selection evaluator, which may improve the prediction rate. Therefore, this research work would be helpful to establish the improvement in accuracy, which might be appreciated by the medical practitioners. In this research work, different tree based classifiers like ADTree, Functional Tree (FT), and Logistic Model Tree (LMT) Classifiers were examined and compared with and without applying Correlation-based Feature Selection (CFS) subset evaluation and the results are analyzed. For this research work, the heart disease data set [37] is taken for analysis. Various measures are used for comparison. After comparison, it is revealed that all the classifiers with CFS attribute selection evaluator has better or equal accuracy as compared without CFS attribute selection evaluator (CFSAE). Top Keywords Heart Disease Prediction, Medical Data Mining, Performance Analysis, Tree-based classifiers, CFS Evaluator. Top |