A Recognition Of Conditional Comment Of Audit By Using Probable Nervous Networks In Companies Listed on Tehran Stock Exchange

Recognition of conditional comment of audit by using methods based on artificial intelligence (Probabilities neural networks) are considered in the research. For this reason, a sample with 1741 observations about financial statements and auditing reports of companies listed on Tehran stock Exchange is collected during 1388 to 1392. Twenty independent variables are used to make probabilities neural network. The maximum variables are extracted from financial statements of companies (financial ralios). Also in order to compare the result of probabilities neural network, the logestic regression model is used. The results showed that probabilities neural network with 91% accuracy has better performance than logestic regression with 89% accuracy in recognition of conditional comment of auditing. These results are correct in observation with accepted comments too. The other part of findings show that variables of intervals. Of the end of financial year and date of auditing report, net sales and ratio of assets carrent have the high importance in recognition of conditional comment of audit.

Keywords:

Auditing , Logestic regression , Probable nervous network , conditional audit report