Abstract

Rice is crucial in Malaysian diet, and Malaysia only produces 4.3 t/Ha of what it needs to support itself. The increasing population of Malaysia requires further increase in rice production for consumption of the country. In recent years, the Malaysian government has been trying to encourage rice production by giving subsidies to the farmers. Accurate forecasting of rice production can provide useful information for the government, planners, decision- and policy makers. The purpose of this paper is to compare between two forecasting methods for rice production estimates in Kedah, Malaysia. The two methods considered and applied on the 35 yearly rice production data are Modified Approach Fuzzy Time Series and Artificial Neural Network. Alyuda NeuroIntelligence software is used for the Artificial Neural Network forecasting. The best model can be determined based on minimum value of mean square error, root-mean-squared error and mean absolute per cent error. The findings of this study indicate that Artificial Neural Network is the best model to be used to forecast rice production in Kedah, Malaysia.