基于RS与ANN的上市公司财务困境预测模型的实证研究

The Prediction Model of Financial Distress of Listed Manufacturing Based on RS and ANN

本文以中国上市公司作为研究对象,采用粗糙集理论(RS)客观选出预测模型指标体系,以因财务状况异常而被列为特别处理公司(ST公司)作为界定上市公司的财务困境标志,采用人工神经网络(ANN)寻找最佳的利用公开财务数据预测中国上市公司财务困境的模型。我们的研究结果表明,总资产报酬率等18个包括现金流量类指标的财务指标有较强的区分财务困境公司和财务健康公司的能力;行业类型和资产规模对于上市公司财务困境预测具有至关重要的作用;运用ANN建立的神经网络模型有较强和较稳定的预测能力。

This paper takes all of the listed companies in manufacturing on Shanghai and Shenzhen stock exchange as the population. We acquire 20 indexes to predict the possibility of financial distresss of listed companies with the theory of attribute reduction and information entropy of rough sets. With ANN, we acquire the prediction model on financial distress of listed companies in manufacturing.