Fuzzy Approaches Ability and their Performance Comparison to Fraud Detection in Financial Reporting

Abstract

The possibility of fraud in the issued financial statements, and its negative impacts on financial markets and the resulting reduction of investment have caused responsible monitoring organizations to detect the frauds and to move seriously against them. This study aimed to investigate the ability of the fuzzy approaches to fraud detection in financial reporting of the firms in the Tehran Stock Exchange. In this study, three hypotheses were considered. First, the Fuzzy decision tree classifier can detect fraud in financial reporting. Secondly, the Sugeno fuzzy classifier can detect fraud in financial reporting. Thirdly, there is a significant difference between the results of fuzzy decision tree classifier and Sugeno fuzzy classifier. These fuzzy approaches were programmed and used for testing the above hypotheses, using Matlab Software. The average accuracy of the Fuzzy decision tree classifier was 31/312, and of the Sugeno fuzzy classifier was 80/92. In other words, the first hypothesis was rejected and the second and third hypotheses were verified.

Keywords

• fuzzy approach
• fraud
• financial reporting
• financial ratio