ENSEMBLED DECISION TREE CLASSIFIER PERFORMANCE WITH VARYING COMMITTEE SIZES

Data Mining is the process of extracting hidden knowledge from the large data repositories. Decision Tree, an important Classification technique of Data Mining is proved its role in medical diagnosis. Ensemble of classifiers will perform better than a single classifier. Prediction error can be minimised by the increase in committee size of classifiers. In this study experiments are conducted on a hybrid Classifier ie., C4.5, a Decision Tree classifier with MultiBoostAB, an ensemble technique with varied committee sizes on Tumour Datasets and the results are analyzed.