Comparative Analysis of Various Tree Classifier Algorithms for Disease Datasets

摘要

TreeBased Classification technique is one of the commonly used techniques called White box classification. It targets foreseeing to the having a place of cases or articles in the classes of a particular variable from their estimations on at least one prescient factor. This research work analyzes the concert of five treebased classification algorithms, namely Decision Stump, J48, Logistic Model Tress (LMT), Random Forest, and REPTree. Various disease datasets such as breast cancer, Pima diabetes, and hypothyroid are utilized for calculating the performance of the classification algorithms by applying the 10fold crossvalidation parameter based on the given class label. Finally, the comparative analysis is held out, using the classification accuracy, kappa value, performance factors, and the error rate measures on all of the algorithms. From the experimental outcomes, it is derived that the LMT provides better results for all the disease datasets than the existing algorithms such as Decision Stump, J48, Random Forest, and REPTree.