پيش بيني قيمت محصولات منتخب كشاورزي ايران با روش تلفيقيِ شبكه عصبي- خودرگرسيوني با ورودي هاي برونزا (NNARX)

Forecasting Iranʹs Selected Agricultural Products Retail Price Using Neural Network Auto-Regressive model with exogenous inputs (NNARX)

In this study, the Neural Network Auto-Regressive model with exogenous inputs (NNARX) was used to forecast the three perspectives ahead of Iranʹs rice, poultry and egg retail price, and to compare its performance with ARIMA model as the most common linear forecasting method. In order to achieve this purpose, the weekly data collected from the Iran State Livestock Affairs Logistics and Countryʹs Welfare Supermarket (related to the period 2002:3-2008:6), and performance evaluation measures such as; R2, MAD and RMSE were used. The results of models performance evaluation showed that the non-linear neural network-autoregressive NNARX model outperforms the linear ARIMA model for all three forecasts and agricultural products.