Perbandingan Akurasi Peramalan Beras di Kota Jambi Menggunakan Metode Dekomposisi, Moving Average, dan Regresi Linier

Abstract

Rice is a strategic food commodity that plays a vital role in community food security. As a hub for government, trade, and services in Jambi Province, Jambi City experiences relatively high rice consumption, whereas local production capacity is limited. This situation necessitates accurate planning of rice requirements through forecasting. This study aims to analyze rice demand in Jambi City, forecast demand using decomposition and regression methods, and compare the accuracy of these two forecasting techniques. The research was conducted in Jambi City in February to July 2026, utilizing secondary data consisting of monthly figures for rice demand and prices from January 2023 to June 2025. Analysis was performed using time-series decomposition and simple linear regression, with forecast accuracy evaluated based on error values. The results indicate that rice demand in Jambi City remained relatively stable in 2023 and 2024, averaging approximately 4,662 tons and 4,642 tons per month, respectively; however, a decline occurred between January and June 2025, with demand averaging around 3,630 tons per month. The decomposition method generated demand forecasts based on trend and seasonal components, whereas the regression method utilized rice prices as an explanatory variable. A comparison of accuracy revealed that the regression method yielded a Mean Absolute Error (MAE) of 298.67 tons, which was lower than the 316.61 tons recorded for the decomposition method. Overall, the regression method proved to be more accurate for forecasting rice demand in Jambi City. These findings can serve as a basis for planning rice availability, procurement, and distribution in the city.