基于舞弊三角理论的财务舞弊识别模型研究——支持向量机与Logistic回归的耦合实证分析

Study of Identification and Countermeasure of Financial Frauds Based on Triangle Theory of Frauds——Empirical Analysis of the Coupling of Machine Learning and Logistic Regression

文章基于舞弊三角理论,对舞弊风险因素进行T检验,选取显著特征运用支持向量机算法和Logistic回归分析方法构建了舞弊识别模型,并进行比较。研究结果表明,财务稳定性越差、监督部门的监督积极性越低、曾经获得非标准审计意见次数越多,公司发生财务舞弊可能性越高。应制定合理的报酬激励机制,降低报酬契约冲突带来的舞弊压力;建立健全内部控制和风险管理评价,减少舞弊的机会;加强内外监督机制,杜绝舞弊的借口。

This paper analyzed fraud risk factors with T-test based on triangle theory of frauds and selected re-markable factors to model the identification of frauds by support vector machine and logistic regression .The re-sults showed that frauds’ increased with the number of the non-standard audit opinions ,but decreased with the financial stability and the positivity of the supervision .In order to prevent and find frauds ,some measures should be taken as follow s .Firstly ,the pressure caused by the conflict of compensation contracts can be re-duced by making reasonable incentive mechanism .Secondly ,the opportunities of frauds can be reduced by im-proving internal controls and evaluation system of risk management .Finally ,the internal and external supervi-sion mechanism can be strengthened to eliminate the excuses of frauds .