Detection of fraudulent transactions using artificial neural networks and decision tree methods

Abstract

The accounting systems generate a large amount of data due to financial transactions. Intentionally fraudulent transactions can occur in high-dimensional and large numbers of emerging data. While many methods can be used for the estimation and detection of fraudulent transactions in accounting, which differ in the audit process, scope and application method, data mining methods can also be used today due to a large number of data and the desire not to narrow the scope of the audit. This study tested the accuracy of detecting fraudulent transactions using artificial neural networks and decision tree methods. According to the results of the analysis test data set for detecting fraud or error risk, 99.7981% accuracy was obtained in the artificial neural networks method and 99.9899% in the decision tree method.