<![CDATA[Uncontrolled urban growth in medium-sized Indonesian cities like Lhokseumawe has created spatial disparities through uneven settlement distribution and numerous underutilized vacant lands, demanding adaptive and data-driven planning approaches. This study aims to analyze spatial patterns of settlement and vacant land distribution in Lhokseumawe City by implementing the Mean Shift Clustering algorithm and to develop spatial planning recommendations supporting sustainable development. The research employs a quantitative approach using spatial data from OpenStreetMap (OSM), Sentinel-2 satellite imagery (for NDVI calculation), and Detailed Spatial Planning (RDTR) data. The non-parametric, density-based Mean Shift Clustering algorithm was applied to adaptively group areas without predetermined cluster numbers, analyzing both city-wide and Blang Mangat District levels. Results show Mean Shift successfully identified 62 spatial clusters across Lhokseumawe City and 28 clusters (OSM-based) plus 37 clusters (RDTR-based) in Blang Mangat District. Analysis revealed linear-transitional urban development patterns from the west (dense urban core) to northeast (transition areas and vacant land), identifying zones of dense settlements, transition zones potential for infill development, and vacant land suitable for green spaces or planned development. Findings also revealed discrepancies between factual spatial patterns from clustering and RDTR zoning plans. The study concludes that Mean Shift Clustering effectively reveals natural spatial structures of settlements and vacant land, providing credible data for spatial policy revision with main recommendations focusing on vertical development, transportation system development along main corridors, and conservation area designation in eastern and northern zones to support Lhokseumawe's sustainable urban planning.]]>
