Mining Financial Statement Fraud: An Analysis of Some Experimental Issues
This addresses the problem of improving fraud detection for financial analysts, but it is incremental as it focuses on analyzing existing experimental issues rather than introducing a new method.
The paper tackles the problem of financial statement fraud detection by analyzing three key experimental issues—problem representation, feature selection, and choice of performance metrics—that influence algorithm performance, critiquing prevailing ideas and providing new understandings.
Financial statement fraud detection is an important problem with a number of design aspects to consider. Issues such as (i) problem representation, (ii) feature selection, and (iii) choice of performance metrics all influence the perceived performance of detection algorithms. Efficient implementation of financial fraud detection methods relies on a clear understanding of these issues. In this paper we present an analysis of the three key experimental issues associated with financial statement fraud detection, critiquing the prevailing ideas and providing new understandings.