Forecasting of rice production in Malaysia using naïve, holt’s linear trend, ARIMA, and feedforward backpropagation neural network

Rice is a vital dietary staple across the globe, especially in Asia. In Malaysia, per-capita rice consumption tops the world average of 54.6 kg, given by the OECD–FAO Agricultural Outlook. According to the Malaysian Adult Nutrition Survey (MANS) findings, Malaysian adults consume an average of two and a half plates of rice daily. Continuing population growth brings about an increase in demand, and consequently, national rice consumption is expected to grow steadily, which may further compel the country to resort to increasing rice imports to satisfy future demand. Yet, this situation might prove adverse for imports due to limitations in the domestic rice industry and depreciation in the Malaysian ringgit, therefore further inducing food security issues. Consequently, pertinent to this study is the immediate need to make an accurate forecast on rice production, so as to feed into strategic planning and policy-making. Different forecasting methods are evaluated in this study using various accuracy metrics, namely the Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), Theil’s Inequality Coefficient (TIC), and Holt’s Linear Trend (HLT), where lower values indicate better predictive accuracy. The analysis also compares conventional with more sophisticated models, which include the Naïve approach, the ARIMA (Autoregressive Integrated Moving Average), and the FBNN (Feedforward Backpropagation Neural Network). A comparative analysis presented a set of strengths and weaknesses across models to highlight their appropriateness for the forecasting of rice production at the national level. The findings drive the need to use methods that are well-developed and primarily data-oriented to increase prediction accuracies. Hence, future works can delve into the concept of hybrid modeling and use more accuracy measures to better the certainty of forecasting. In the grand scheme, this study carries a lot of weight in agricultural forecasting in Malaysia, aiding decision-making processes for producers, stakeholders, and policymakers, thereby guaranteeing food security in the long term.