Authors

• Iman Farjamnia ¹
• Mohsen Naseri ²
• Sayed Mohamad Mehdi Ahmadi ³

¹ M.A in Energy Economics

² Ph.D Student in Civil Engineering, Shiraz University

³ Graduate Student, University of Tehran

Abstract

Ability of Artificial Neural Networks (ANNs) as a powerful tool in simulation and prediction in science and engineering has made it attractive to economists. In this article, after a brief review of literature, a comparison of forecasting performance of ANN versus ARIMA is made. The data used are daily prices of oil for the period April 1983 to June 2005. In addition, sensitivity analysis is implemented for illustrating contribution of each input to the price changes in ANN models. The results show that the ANN model generates more accurate forecasts for the daily oil prices of oil than ARIMA model.

Keywords

• Time series
• Artificial Neural Networks (ANN)
• ARIMA
• Sensitivity Analysis