Using of Self-Organizing Maps for Display and Analysis ofMultidimensional Spaces In Accounting

This research deals with the comparative examination of applying artificial neural networks and auto regressionmodel and its influence in declining prediction error of Tehran’s stock market index. The basic subject of theresearch is to find a method to the better prediction of stock price. Statistical societies of this study are thosecompanies accepted in Tehran’s stock market during 2002 –2014. Statically sample of the research equals to thestatistical society .Required data are gathered by library studies for research literature and by country stock marketorganization software’s for variables. According to the achieved results of comparing the actual performance withprediction by the regression line and prediction by artificial neural network, it can be concluded that the differencebetween those predicted by artificial neural network and actual performance is less than the difference betweenthose predicted by regression line. This may be due to the nonlinear nature of neural network, and also randomsteps theory and performing of that in this kind of network

کلیدواژه ها:

Artificial Neural Networks ، Auto regression ، Exchange index ، Predictor