Comparison of performance of fuzzy autoregressive integrated moving average and fuzzy neural network to forecast economic growth in Iran

Since multivariate econometric forecasting encounters with much restriction، using univariate models seems to be a proper alternative approach. But most of these methods require large amounts of data to achieve a good result. Fuzzy regression approach which uses the fuzzy numbers for modeling and prediction needs less data compare to nonfuzzy approaches. In this study، fuzzy autoregressive integrated moving average (FARIMA)، which combines autoregressive integrated moving average (ARIMA) with fuzzy regression has been used. The forecasting performance of this model has been compared with prediction performance of ARIMA and adaptive neural-fuzzy inference system (ANFIS) methods to forecast economic growth in Iran during the period 1960 to 2002. The forecasting performances of these models have then been compared to forecast economic growth for the period 2003 to 2010 using the criteria of RMSE، MAE، MAPE and TIC. The results of the study showed that FARIMA model has the best performance، and ANFIS method suggests better forecasting that ARIMA.