Application of Random Forest, Rough Set Theory, Decision Tree and Neural Network to Detect Financial Statement Fraud – Taking Corporate Governance into Consideration

Abstract

AbstractGiven that corporate financial statement fraud cases have been increasing in recent years, establishment of a model to effectively detect financial statement fraud has turned out to be one of the crucial issues. This study has integrated RF, RST, decision tree and back propagation network (BNP) to construct a corporate financial statement fraud detection model, so as to help auditor detect the signs of financial statement fraud before a fraud occurs in an enterprise. The study has established a new financial statement fraud detection model, for which the empirical result shows that, while analyzing the corporate financial statement fraud indexes considered and measured by RF, corporate governance indexes should also be included in financial statement fraud detection other than other financial indexes. In addition, it is also found that the corporate financial statement fraud detection model constructed by integrating RF and RST could also effectively enhance classification accuracy.KeywordsRandom Forest(RF)Rough Set Theory(RST)Decision TreeNeural NetworksFinancial Statement FraudCorporate Governance