A Road Map for Detecting Financial Frauds in Enterprises through Data Mining

Abstract

Enterprises face the problems of combating financial frauds in their various locations. Although, most of these enterprises engage full time internal auditors, the manual process carried out particularly to detect frauds, have not been completely assured. This is as a result of the large volume of financial transactions existing such that only random sampling of the audit is presently done.

A financial fraud detection is efficiently conducted if the process successfully passes through journal entries as the primary source of financial statements.

This road map having reviewed the previous attempts of researches such as Benford's law, Neural networks and Self organizing maps to detect financial frauds as modest contributions, describes a data mining process using decision tree algorithm that builds pattern models. An SQL Server Integration Services (SSIS) are utilized to affect the decision tree data mining to reduce anticipated frauds and enables a complete auditing. This resulted in providing complete, accurate, confident and reliable audit results to the enterprise managers.

The Journal Entries data mining revealed patterns of attempted frauds in enterprises particularly patterns gleaming unapproved entries, entries of amounts above approval limits, entries made during holidays and weekends