Analisis Pola Produksi Pangan di Pulau Sumatera Berdasarkan Data Time-Series dengan Model Autoregressive Integrated Moving Average (ARIMA) dan Metode Optimasi Grid Search

Abstract

Food security is a critical concept defined by FAO as the ability to ensure access to sufficient, safe and nutritious food for every individual. Sumatra Island, which ranks second in Indonesia's Food Security Index in 2022, faces complex challenges in maintaining food security in the aspects of availability, accessibility, utilization, and stability. This study focuses on analyzing food production patterns in Sumatra through time-series data modeling on three main commodities: rice, corn, and soybean. The ARIMA (AutoRegressive Integrated Moving Average) method was optimized using grid-search to identify trends and seasonal variations in the 1998-2022 time span. The results showed an error rate of RMSE (root of MSE) with an average of 99129.30 and MAPE of 60.02%. Coefficient of Variation (CV) analysis revealed a significant increase, for the first year the average was 1868.54%, while for the last year it was 5503.56%. Data visualization presents a clear pattern of production trends, providing strategic insights for stakeholders. Model validation through stationarity tests, differencing, and evaluation metrics ensure forecasting accuracy. This research provides a comprehensive analytical foundation for food security policy planning in Sumatra.