Forecasting seafood production in Iran using the ARIMA and Artificial Neural N.etworks

The main objective of this paper is to predict the value of seafood production in Iran. For this purpose, collective self-explanatory time series moving average (ARIMA) and artificial neural networks are used. In this study, three different structures of neural networks, including multilayer forward neural network, radial basis function and Elman are used. Fisheries organization publishes the seafood production rate yearly for research purposes. In this study data of 1995 to 1911were utilized. Then, the yearly data converted to monthly by using statistical software. The results showed that based on MAPE, the maximum error cost is 0.0771 for ARIMA and the minimum error cost is 7.9328 × 10-5 for RBF neural network. It is concluded that RBFwas most accurate method to predict seafood production for the next two years.