Using Decision Tree Algorithms and Artificial Intelligence to Increase Audit Quality: A Data-Based Approach to Predicting Financial Risks

Abstract

The use of AI has increased significantly in most professions, and the auditing profession is no exception. AI systems have significantly changed the auditing process. However, opponents of the AI ​​revolution, such as many auditors who have not adapted to the new changes, see this development as a step backwards. Meanwhile, the main issue at the beginning is the perception of auditors regarding the function of AI in audit quality, because auditors' perception of AI plays an important role in its use by auditors. Auditors' perception depends on important variables such as perceived convenience and perceived usefulness. Therefore, the main objective of this research is to analyze auditors' perception of AI and its contribution to audit quality. To predict audit quality using decision tree algorithms. Therefore, all auditing firms that are members of the Iranian Certified Public Accountants Association during the period 2008-2024 constitute the statistical population of the research, and 4367 observations remain as a statistical sample after screening. This research is applied in terms of purpose and descriptive in terms of research method. Data analysis was performed by applying CRISP-DM data mining standard and four decision tree algorithms, namely CHAID, C&RT and C5.0, and QUEST. The results showed that regardless of the depth of the tree, the optimum models with the highest detection power of 98% were associated with the C5.0 tree and more than 93% with the C&RT tree. Therefore, out of the total 19 audit quality assessment criteria, 16 criteria in C5.0 algorithm, 12 criteria in CHAID algorithm, 5 criteria in C&RT and 3 criteria in QUEST were considered effective in predicting audit quality, and the rest were eliminated. It is important to note that the common criteria in the four algorithms, namely employee recruitment, employee training and job control, and audit planning, are the input stages affecting audit quality.