This paper presents the results obtained from the compar- ison of three different methods for creating an ensemble of classifiers applied on particular problems taken over from the University of Cal- ifornia, Irvine machine learning repository. All data sets deal with the binary classification problem. On each of the data sets boosting, bag- ging and random forests methods for generating ensemble of classifiers were applied. Results from the generated ensembles were used to make comparison between different methods applied. The idea was to find the best method among these three, if possible, and generalize the conclusion for other data sets with similar characteristics as those used. Decision stumps or pruned or unpruned decision trees were used as base classi- fiers for boosting and bagging methods. There was also a variation in the number of classifiers used in the ensemble.
Ensemble, Boosting, Bagging, Random Forest