Application of decision tree algorithms in predicting audit quality

The purpose of this paper is to predict the quality of the audit using decision tree algorithms. Therefore, all audit institutions of the member of the Iranian Society of Official Accountants during the period 1391 to 1396 are the statistical population of the study, which after the screening of 1367 observations remained as a statistical sample. This research is a descriptive-correlative research in terms of applied and descriptive research method. Data analysis was performed in accordance with the CRISP-DM data mining standard and the implementation of four decision tree CHAID, C & RT, C5.0, and QUEST algorithms. Decision trees were modeled using simulation software of IBM modeler18 and the results showed that the optimal models, regardless of tree depth, with the maximum recognition power associated with the tree C5.0 over 97% and considering the depth of the tree with more than 92% is related to the C & RT tree. From the total of 19 quality assessment criteria, 16 criteria in the C5.0.12 Criterion Algorithm in the CHAID algorithm and 5 C & RT criteria and 3 criteria in QUEST are considered in the prediction of effective audit quality It is important to note that the common criteria in all four algorithms, which are employee recruitment, employee training, and entrepreneurship All work control and supervision are all input phases that affect audit quality.