Investigating determinants of disclosure quality using artificial neural network

The purpose of this research is to propose a model for predicting disclosure quality using artificial neural network. Toward this end, this research has used the variables related to liquidity, profitability, leverage, company size, corporate governance and other effective variables by using the artificial neural network. Minimal-Redundancy-Maximal-Relevance criterion and sequential feature selection as two Feature selection methods are used to preprocess the data that could improve the accuracy of model. Results show that in the model where all variables are applied, the linear regression correlation between network output and scope data is %87.8 while in the model where the seven variables of Ownership concentration, Assets-in-place, Age, Profit margin, Percentage of non-executive board members, Institutional ownership ratio, and Number of employees are used, this correlation stands at %92.26. Also, these results show the significant effect of the corporate governance variables on disclosure quality.