<![CDATA[The high crime rate in DKI Jakarta requires spatial analysis to accurately identify vulnerable zones. Such information is essential for developing data-driven crime prevention strategies. Therefore, this study aims to map the spatial distribution of criminal cases in DKI Jakarta in 2024 and to evaluate two clustering methods Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and K-Means in order to determine the most effective approach for identifying crime-prone areas. Data came from the Jakarta Open Data portal, containing coordinates (latitude, longitude), crime types, and supporting details. Pre-processing involved removing duplicates, filtering 2024 records, and eliminating invalid coordinates. Spatial features were normalized using Standard Scaler. DBSCAN parameters (eps, min_samples) were tuned via grid search, while K-Means used the Elbow method to determine optimal clusters. Performance was evaluated using Silhouette Score, Davies-Bouldin Index, and Calinski-Harabasz Index. K-Means achieved a higher Silhouette Score (0.483), lower Davies-Bouldin Index (0.725), and higher Calinski-Harabasz Index (145.897), indicating more compact, well-separated clusters. DBSCAN formed more clusters (12) and identified noise points, showing its ability to capture spatial density variations and detect small-scale hotspots. In conclusion, K-Means is more suitable for macro-level mapping of security areas, supporting the allocation of police resources and administrative planning, while DBSCAN is more effective in identifying localized hotspots that require targeted surveillance and rapid response. These findings provide practical insights for policymakers and law enforcement agencies in developing data-driven strategies for urban crime prevention and security management.]]>
Copyrights © 2026
