The Performance of Single Classifier, Ensemble without Diversity and Ensemble with Diversity

The performance of ensemble depends on the single classifiers chosen. Diversity in ensemble could be a factor that may influence the results or the performance of ensemble. In this study we have employed bagging and boosting as ensemble classifier, DECORATE to tackle diversity in ensemble. We have chosen random forest, random tree, j48 and j48 grafts mainly as a base classifier for the ensemble methods. The empirical evidence has shown that Boosting algorithm without diversity do not improve the test performance of the single classifier.