Abstract:
In this study, application of Adaptive Neuro-Fuzzy
Inference System (ANFIS) in
forecasting three perspectives (1, 2, and 4 years)
ahead of Iran’s agricultural products
export was compared with ARIMA as the most common e
conometrics linear forecasting
method. For this purpose, Iran’s agricultural produ
cts export revenues related to 1959-
2010, and forecast performance measures such as R
2
,
MAD, and RMSE were used.
Results of the models performance evaluation showed
that the forecasted test data related
to ANFIS designed architects had more correspondenc
e with the real data in comparison
with that of ARIMA forecasted out of sample data. T
herefore, the non-linear ANFIS
model outperformed the linear ARIMA model for all o
f the considered perspectives.
Keywords:
ANFIS, ARIMA,
Non-linear
.
