Affiliations:
College of Computing, Informatics and Mathematics, Universiti Teknologi MARA Cawangan Negeri Sembilan, Kampus Seremban, Negeri Sembilan 72000, Malaysia
This study aims to forecast future rice production in Malaysia concerning national targets by comparing the effectiveness of three models: Autoregressive Integrated Moving Average (ARIMA), Exponential Smoothing, and Long Short-Term Memory (LSTM). Unlike ARIMA and Exponential Smoothing, which are based on predefined statistical assumptions, LSTM uses deep learning to detect complex, non-linear, and long-term patterns in time series data. The performance of these models, applied to Malaysia’s annual rice production data from 1960 to 2023, was evaluated using error measures such as Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). Results showed that Double Exponential Smoothing produced the lowest error rates, making it the most accurate method for predicting rice production. While LSTM is considered a more advanced technique, it did not perform better than Double Exponential Smoothing in this case. The study concludes that predicted rice production levels are likely to fall below government targets over the next five years. This finding emphasizes the need to focus on sustainability strategies, such as reducing reliance on imports and enhancing domestic rice production. The results can guide policymakers in addressing future challenges, promoting sustainable agricultural practices, and ensuring Malaysia's long-term food security. Future research could explore using hybrid models in a multivariate setting and expanding datasets to compare regional and global rice production trends.
Rice production, Forecasting models, Agricultural practices, Food security, Sustainability strategies
https://doi.org/10.21833/ijaas.2025.09.016
Shafie, N. A., Rosli, A. R., & Azmi, W. N. S. W. (2025). Modeling and forecasting Malaysian rice production: Insights from ARIMA, exponential smoothing, and LSTM models. International Journal of Advanced and Applied Sciences, 12(9), 169–179. https://doi.org/10.21833/ijaas.2025.09.016