<![CDATA[Climate variability poses a major challenge to rice production in Sumatra, a key contributor to Indonesia’s food security. This study aims to analyze spatiotemporal climate impacts on rice yields by integrating climatic, geographical, and agricultural datasets. Historical records from 1993–2024, including rainfall, temperature, humidity, and rice production statistics, were collected from BMKG, BPS, and the Ministry of Agriculture. After preprocessing and feature selection, six machine learning algorithms—Linear Regression, Random Forest, Gradient Boosting, Support Vector Regression, Decision Tree, and K-Nearest Neighbors—were evaluated for predictive performance. Results show significant spatial heterogeneity: rainfall strongly affects yields in Aceh and North Sumatra, while temperature stress is critical in southern provinces. Among the tested models, Random Forest achieved the best accuracy (R² = 0.985), outperforming other algorithms. These findings highlight the importance of localized adaptation strategies and demonstrate the potential of ensemble machine learning to support climate-resilient rice production.]]>
Copyrights © 2025
