1Faculty of Management and Accounting, Univ. of Sistan and Baluchestan, Zahedan, Iran
2Faculty of Engineering, Shahid Bahonar University, Kerman, Iran
3Faculty of Management and Accounting, University of Sistan and Baluchestan, Zahedan, Iran
Online published on 13 February, 2012.
The nature of descriptive relations in most of real life processes, especially in commercial and managerial fields, is mostly nonlinear. Therefore, behavior forecasting of such processes needs accurate and effective tools.Artificial neural networks, as important model- making. Tools in forecasting business issues are able to compensate the deficiencies of common models. This paper’s aim is to demonstrate the superiority of neural networks in forecasting nonlinear processes to other forecasting models. To fulfill this aim in this paper Iran’s wood industry data, including production volume, import volume and currency-based value of importation from1961 to 2007, have been studied. At first by the usage of these data, neural network and nonlinear models implementation; which derived from MATLAB software, some forecastings were made about Iran’s wood industry and then based on MAPE the yielded results of the aforementioned approaches were compared with each other. Investigation results revealed the outstanding success of the neural network in all three studies in comparison to nonlinear models derived from MATLAB software.
Forecasting, neural network, nonlinear processes
