Using data mining techniques in audit

Investors and decision makers have to believe that the information is reliable and transparent since they do not have the opportunity to research directly at the source of the information presented to them. üFinancial statements that have not been independently audited in line with national and international standards will have a high probability of containing errors and fraud. For this reason, investors and decision makers want to have an objective assessment of the financial situation of the organization. The need for transparent and reliable financial statements has revealed the need for independent auditing. An independent audit provides reasonable assurance that the entity's financial statements are prepared in accordance with the applicable financial reporting framework and increases transparency in the financial reporting process. Artificial neural networks are a sub-branch of artificial intelligence that is used to analyze complex information and is one of the qualitative methods. Networks have great potential to solve problems that cannot be solved with logical and analytical techniques with standard software. Artificial intelligence (AI) is rapidly changing the way financial institutions work. AI is expected to increasingly take over the core functions of the business due to cost savings and operational efficiencies. The aim of this study is to explain the advantages of using the artificial neural network model in the planning phase of the audit. For this purpose, an artificial neural network model was created. While creating this model, 8 neurons 10,000 iterations were used in the 2-layer network structure. As a result of various trials, the network structure with the best estimation performance was preferred. The application part of the study consists of joint stock companies traded in the BIST-Spor index. The data set of the study consists of the financial information obtained from the financial reports of the companies published on the Public Disclosure Platform. It is based on the data in the financial statements of the companies between 2013-2018. Using the artificial neural network model, one of the data mining techniques used in independent auditing, the data for 2019 were tried to be estimated. In this context, the years 2013-2016 were used as the training set and the years 2017-2018 as the test set. Calculations for 2019 were estimated by creating a model from 2 layers of 8 neurons. In the analysis made, it has been observed that the difference between the value estimated by the auditor and the actual value is less in some accounts and more in others. Thus, the auditor will be able to focus on the accounts where the difference is large while performing the planning. The MAPE value was calculated from the estimation performance measurements. Accordingly, while a detailed examination is made for accounts with a MAPE value above 20%; Accounts with a value below 10% can be ignored. In this way, the auditor will save time by completing the audit in a shorter time within the framework of risk assessment procedures, audit risk will be reduced and audit quality will increase.