Two-step classification method based on genetic algorithm for bankruptcy forecasting

作者：

Highlights：

• A method of design of voting ensemble of heterogeneous classifiers is proposed.

• The method combines feature selection and random sampling techniques.

• Results confirm the ability of method to select the task-relevant features.

• Factors of external environment are very important for quality of bankruptcy forecasting.

摘要

•A method of design of voting ensemble of heterogeneous classifiers is proposed.•The method combines feature selection and random sampling techniques.•Results confirm the ability of method to select the task-relevant features.•Factors of external environment are very important for quality of bankruptcy forecasting.

论文关键词：Bankruptcy,Bankruptcy forecasting models,Ensembles of classifiers,Features selection,Genetic algorithm

论文评审过程：Received 9 November 2016, Revised 16 June 2017, Accepted 15 July 2017, Available online 17 July 2017, Version of Record 20 July 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.07.025