<![CDATA[Rice price volatility significantly impacts economic stability and food security in Indonesia, particularly in East Java, where fluctuations in staple food prices affect household purchasing power and inflation management. This study addresses the limitations of existing rice price forecasting models, which often struggle to capture the complex, nonlinear dynamics of agricultural prices influenced by multiple factors such as climate variability and market conditions. Accurate and reliable price forecasting is essential to support effective policy formulation, market intervention, and food price stabilization strategies. This research develops an ensemble forecasting framework integrating Gated Recurrent Unit (GRU) and Support Vector Regression (SVR) with enhanced feature engineering to predict daily medium rice prices using historical price and weather data. The dataset comprises daily observations from 2021 to 2025, including rice prices, average temperature, relative humidity, rainfall, and sunshine duration. In this framework, GRU serves as a temporal feature extractor to learn complex temporal dependencies, while enhanced feature engineering generates complementary statistical features from sliding windows to enrich GRU's output. The combined feature set is provided to an SVR model with a Radial Basis Function kernel for final regression. Experimental results show that the proposed model achieves a high forecasting accuracy with an MAPE of 0.109%, demonstrating stable predictive behavior and making it a valuable tool for monitoring rice prices. The model's effectiveness in capturing temporal dependencies and nonlinear patterns suggests potential applicability beyond East Java, offering broader insights for agricultural price forecasting in other regions.]]>
