Short-Term Rice Yield Forecasting in Pangasinan: A Seasonal Autoregressive Integrated Moving Average (SARIMA) Modeling Approach

Abstract

Rice farming in Pangasinan is critical to Philippine food security but increasingly vulnerable to climate volatility. Traditional intuition-based farming calendars are insufficient against erratic weather patterns, leaving stakeholders with reactive rather than anticipatory strategies. This study aims to develop a short-term forecasting model to empower datadriven decision-making. Utilizing quarterly rice yield data from 2008 to 2025, the study employed the Box-Jenkins methodology to formulate a Seasonal Autoregressive Integrated Moving Average (SARIMA) model. The analysis revealed a robust "sawtooth" seasonal pattern, where the fourth quarter consistently yields peak production, while the third quarter suffers a significant lean season. Following rigorous stationarity testing and differencing, the SARIMA (0,1,1) (0,1,1)4 model was identified as the optimal predictor (AICc: 1575.88), outperforming automated selection algorithms. Diagnostic checking confirmed the model successfully captured deterministic patterns. Projections for late 2025 forecast a robust yield of 751,139.8 MT. However, the model explicitly quantifies a critical vulnerability in the third quarter of 2026, projecting a precipitous drop to 112,620.2 MT or 85% contraction from peak capacity. These findings provide agricultural authorities with the foresight needed to optimize resource allocation and mitigate the economic risks of seasonal scarcity.