A novel bagging C4.5 algorithm based on wrapper feature selection for supporting wise clinical decision making

Shin-Jye Lee, Zhaozhao Xu, Tong Li, Yun Yang

Abstract: Highlights•Our sampling strategy optimally integrates the advantages of random sampling and oversampling techniques.•The wrapper method in making effective feature selection is strongly supported by our sampling strategy without processing huge amounts of data.•Our approach does not only improve the feature selection performance, but also enhances the classification performance of the Bagging C4.5 as a whole.•Extensive experiments show that our approach outperforms all the existing representative benchmarks for supporting wise clinical decision making.