An Improved Oversampling Algorithm Based on the Samples’ Selection Strategy for Classifying Imbalanced Data

Table 3

The comparison of classification results for the different methods.

Data set ID	Original-SVM		SMOTE		Random-SMOTE		AKN-Random-SMOTE
	G-mean	F-measure	G-mean	F-measure	G-mean	F-measure	G-mean	F-measure
1	0.634	0.114	0.698	0.724	0.739	0.762	0.784	0.810
2	0.605	0.451	0.670	0.665	0.750	0.738	0.860	0.864
3	0.653	0.594	0.673	0.661	0.712	0.711	0.812	0.781
4	0.672	0.540	0.677	0.533	0.832	0.839	0.895	0.922
5	0.739	0.664	0.786	0.806	0.787	0.799	0.915	0.890
6	0.257	0.126	0.628	0.563	0.644	0.584	0.725	0.690
7	0.925	0.895	0.933	0.906	0.937	0.935	0.968	0.967
8	0.898	0.790	0.951	0.951	0.974	0.975	0.981	0.981
9	0.946	0.909	0.954	0.956	0.940	0.941	0.973	0.967
10	0.802	0.780	0.964	0.962	0.958	0.957	0.975	0.975