An empirical analysis of decision tree algorithms: Modeling hepatitis data

Data mining refers to the process of retrieving knowledge by discovering patterns from large datasets. This paper highlights the performance of seven decision tree classification algorithms viz. Decision Stump, Hoeffding Tree, J48, Logistic Model Tree(LMT), Random Forest, REP (Reduced Error Pruning) Tree and Random Tree on the Hepatitis prognostic dataset that enables the classifier to accurately carry out categorization of medical data. The classification accuracies are evaluated using 10 fold cross validation technique. The results affirm the fact that the Random Forest algorithm better performs all other algorithms.