Data mining methods used to determine financial statement frauds and an application in Borsa Istanbul

Financial scandals experienced in the world and in our country regarding businesses that have been independent audited and received a reasonable level of assurance have been effective in discussing the accuracy and reliability of the financial reports issued by these companies. Discussion of the data included in the financial table also caused hesitation by the table users in their decisions. In this study, a model has been developed in order to eliminate the hesitation of the table users and to predict the financial table fraud that has occurred or may occur in the financial table at a certain level of assurance. As data in our research; independent audit reports of 144 businesses operating in the close monitoring market, star market and main market groups operating in Borsa Istanbul between 2012-2019 and the data obtained from the financial tables are used. Datas obtained from the financial table of the enterprises, 48 of which are in the close watch market and 96 of which are in the star market and in the main market groups and datas used in the detection of financial table fraud in the literature are analyzed using the method of artificial neural networks, one of the data mining applications, and subsequently an artificial neural network model was developed. The developed model has revealed a successful result by correctly estimating the fraud risk in the financial table at the rate of 88,89%. The results of the research have been evaluated that the developed model will be beneficial in the decisions of information users regarding whether the companies carry the risk of financial report fraud or not. It has been concluded that by using the parameters of the model, it will give a strong assurance about whether there is a risk of financial reporting fraud about the enterprises that information users will invest in and will be beneficial to information users.