基于数据挖掘的上市公司财务舞弊识别研究

国内上市公司财务舞弊呈恶性发展之势,要推进证券市场的预警与监管,对上市公司财务舞弊行为的识别就变得意义重大。以148个财务舞弊上市公司和配对的非舞弊上市公司为样本,运用神经网络建立了上市公司财务舞弊识别模型。此模型对训练样本和测试样本的识别正确率分别达到74.58%和70%,能有效的识别出上市公司财务舞弊与否。结果表明,该模型可以用于上市公司财务舞弊行为识别,对有舞弊动机的上市公司起到威慑作用。

Research on the Financial Fraud Identification of the Listed Companies Based on Data Mining Techniques

Domestic listed companies shows a worsen trend of financial fraud behavior. It becomes significant to recognize the financial fraud of listed companies, in order to promote early warning and supervision of the securities market. 148 samples of financial fraud listed companies and counterparts of Non-financial fraud were collected, on which Neural Network is employed to establish the identification model of financial fraud on Chinese listed companies. The correct rates of this model running on training samples and test samples are 74.58% and 70% respectively, showing this model can effectively identify the financial fraud behavior of listed companies. Conclusion is drawn that the model can be applied to identify financial fraud behavior of listed companies, and can play a deterrent role on listed companies who have fraud motivation.