Empirical Studies and Analysis of Ensemble Learning Techniques in Data Mining

Abstract

This Classification using ensemble generally combines multiple classifiers that results in the improvement in the accuracy of the classification. Experimenting with the same dataset using the single classifier provides lesser accuracy than ensemble techniques. Many researches have been carried out using the technique of combining the predictions of multiple classifiers to generate a single classifier. The produced classifiers provide more accurate results than any individual classifier. This paper focuses on the ability of ensemble techniques to improve the accuracy of basic J48 algorithm. Ensemble techniques like Bagging and Boosting improved the efficiency of the J48 classifier. Experiments have been carried out on many datasets taken from UCI repository to investigate the effects of ensemble techniques on J48 and Naïve Bayes algorithm. WEKA tool is used to measure the effectiveness of a classifier model.