Comparison of Arima Method and Artificial Neural Network Method to Predict Productivity Rice In Panti District

Abstract

Rice production is a community activity to produce rice, it is intended to maintain food security in the future. The aim of this research is to develop the best model for forecasting rice production based on ARIMA (Autoregressive Integrated Moving Averages) and ANN (Artificial Neural Network) approaches. The results will be compared with the error rate values of the ARIMA and ANN methods with the available data. The data used in this study is data on rice production in Panti District, Jember Regency. The level of forecasting accuracy produced by each forecasting method is measured by the criteria of MAPE (Mean Absolute Percentage Error), MSE (Mean Square Error) and RMSE (Root Mean Square Error). The results showed that from the forecasting method used in this study, the ARIMA (1,0,1) (1,0,2) method is the best forecasting method for the best rice harvest area in Panti District, Jember Regency with an average MAPE value is 0.05668374, MSE is 5.587553, and RMSE is 2.3638. Meanwhile, forecasting rice productivity using the ANN BP method (7,(7,3),1) is a fairly good forecasting method with an average MAPE value of 0.05703856 MSE of 4.828465, and RMSE of 2.197377. Therefore, the ARIMA model (1,0,1) (1,0,2)[12] is quite effective for predicting the amount of rice production in Panti District, Jember Regency, East Java Province for the next few years.