Is Hybrid Method Better than ANN Single Method to Forecasting The Price of Indonesia's Rice?

By : Ahmad Jauharudin Najib, Student of S1 Statistika Universitas Muhammadiyah Semarang

The price of rice has uniqueness in the process of determining it, so it is necessary to be careful in determining the price. Therefore, it is necessary to do the appropriate forecasting Research about forecasting employing Autoregressive integrated moving average (ARIMA) and Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) method has been done previously. While other forecasting methods are also proposed, this research consider employing Artificial Neural Network (ANN), a combination of ANN and ARIMAX called Hybrid NNs-ARIMAX. Combining these two methods is done to get more accurate results.

First of all, I think that hybrid method is better than ANN single method to forecasting the price of Indonesia's rice. This is because the ANN method is designed to imitate the workings of the human brain in solving problems. The main advantage of the ANN Method  is the ability to learn from the examples given. While the Arimax method involves more than one variable. Each method has its advantages. 

Hybrid is a method that combines the two methods. In my opinion hybrid method has a higher level of accuracy than the ANN single method. It is because, if the advantages of each method can increase the accuracy of forecasting. So if the  two methods are combined, of course the level of prediction accuracy will also increase.

In additional, Research conducted in 11 provinces in Indonesia, Resulting in the fact that 2 provinces are better off using the NNs-ARIMAX hybrid method, and And 9 other provinces, it is better to using the NNs-ARIMA hybrid method. It can be seen from the increase in forecasting capabilities, when using the Hybrid NNs-ARIMA and Hybrid NNs-ARIMAX methods compared to the ANN single method. The MAPE increasing about 1.21% when using Hybrid NNs-ARIMA, and 0.23% when using Hybrid NNs -ARIMAX. So this proves that the hybrid method is better than the ANN single method.

In conlusion, I agree with the fact that all provinces are more suitable to use the Hybrid method compared to the ANN single method. And look at the indicator of the MAPE value that increases when using the hybrid method. So that hybrid  method is better than ANN single method to forecasting the price of Indonesia's rice.