Igf-bagging: Information Gain Based Feature Selection for Bagging

Bagging is one of the older, simpler and better known ensemble methods. However, the bootstrap sampling strategy in bagging appears to lead to ensembles of low diversity and accuracy compared with other ensemble methods. In this paper, a new variant of bagging, named IGF-Bagging, is proposed. Firstly, this method obtains bootstrap instances. Then, it employs Information Gain (IG) based feature selection technique to identify and remove irrelevant or redundant features. Finally, base learners trained from the new sub data sets are combined via majority voting. Twelve datasets from the UCI Machine Learning Repository are selected to demonstrate the effectiveness and feasibility of the proposed method. Experimental results reveal that IGF-Bagging gets significant improvement of the classification accuracy compared with other six methods.