多源信息融合视角下中小企业财务危机预警研究——基于集成学习的数据挖掘方法

Research on SMEs' Financial Distress Prediction from Multi-source Information Fusion Perspective——Based on Ensemble Machine Learning Method

大数据环境下,影响中小企业经营状况的因素愈发复杂。为更加准确地衡量中小企业的财务危机状况,从多源信息融合视角出发,在传统反映企业私有信息的财务数据基础上,融合分析了衡量宏观经济和市场环境的公共信息数据。考虑到多源信息数据的多维异构性,引入BPNN、SVM、KNN、LOG和MDA建立模型,并运用集成学习对不同预警模型的结果进行集成处理,以降低不同模型结果不一致带来的风险。实证结果证明融合多源信息的预警模型能提高预警准确率。

Under the big data environment,the influencing factors of SMEs' management state is more complex.In order to predict SMEs' financial distress more accurately,the paper considers fusion of new public information which measure macro-economic and market environments and traditional private financial information,from multi-source information fusion perspective.Five classic data mining methods,namely BPNN,SVM.KNN,LOG and MDA,are introduced to process the multi-sourced heterogeneous data.To reduce the risk of inconsistent results,ensemble learning method is involved to integrate the results from the five methods.The empirical results show that the proposed model which integrates multi-source information can significantly enhance the prediction accuracy.