<![CDATA[The classification of 23 regencies/cities in West Java from 2008 to 2024 was executed using the K-Means algorithm on a dataset spanning five variables: production, harvested area, productivity, population, and agricultural workforce. K-Means was chosen for its efficiency and ease of interpretability when analyzing large-scale multivariate data across time. Optimal cluster determination involved evaluating the Elbow Method, Silhouette Score, and the Davies-Bouldin Index (DBI). Although K=5 was suggested by the Elbow Method, K=6 was selected because it demonstrated a more stable and representative regional separation, supported by the lowest DBI of 0.8221 and a relatively high Silhouette Score of 0.4531. Cluster boundaries were further validated through PCA and GIS visualization. The analysis revealed precise regional segmentation. Key findings indicate that Indramayu, Karawang, and Subang regencies are stable, high-production centers, suitable for intensification and modernization. Conversely, regions like Bandung and Garut regencies exhibited dynamic cluster shifts driven by urbanization and climate variability. This segmentation has crucial policy implications: stable areas are suitable for intensification, dynamic areas require adaptive risk-mitigation policies, and urban-influenced regions (Bandung, Bekasi, and Depok cities) must focus on diversification and agricultural innovation. Despite the limitations of K-Means’ inability to capture complex, non-linear clusters, this research highlights the value of integrating spatiotemporal clustering for policy insights. Future research should incorporate climate and land-use data with advanced clustering methods, such as DBSCAN and HDBSCAN. HDBSCAN is more suitable for modeling clusters with varying densities, and time-series approaches should also be integrated. Overall, these results provide an essential, evidence-based framework for targeted agricultural planning.]]>
Copyrights © 2025
