Abstract: This study tends to model Gordon relation using feed forward neural network approach for Tehran Stock Exchange listed companies. In this research, the examination of Gordon model with a nonlinear approach is discussed and the results are compared with linear r More
Abstract: This study tends to model Gordon relation using feed forward neural network approach for Tehran Stock Exchange listed companies. In this research, the examination of Gordon model with a nonlinear approach is discussed and the results are compared with linear regression. The examination of nonlinear Gordon model using neural network has not been considered in the studies as yet. In this research, data for 247 companies and 1135 observations (firm- year) between 2006- 2013 are used (unbalanced panel). Comparative analysis of results for linear regression and nonlinear neural network approaches shows that the coefficient of determination for neural network approach is higher than the coefficient of determination for linear regression; so using nonlinear model can improve prediction power of model and lead to more profitable investment strategies. In the modeling procedure, various structures of network are tested by changing the number of neurons for getting optimal network.
Manuscript profile
