Kernel density and spatial modeling of informal settlement concentration: methodology and findings from Palembang, Indonesia

Rapid urbanization has intensified the growth of slum settlements in Indonesian cities, including Palembang, where informal housing commonly develops along riverbanks. This study aimed to identify and evaluate the spatial distribution and density of slum areas in Palembang City through a Geographic Information System (GIS)–based approach combining Kernel Density Estimation (KDE) and Receiver Operating Characteristic–Area Under Curve (ROC– AUC) analysis. Primary spatial data were obtained from 382 household survey points representing 64 slum polygons across 13 sub-districts, supplemented by administrative boundary and land-use data from the Palembang City Government. Spatial analysis and validation were conducted using ArcGIS 10.3 software. The KDE results showed density values ranging from 0 to 58.1123 units per 100 m², with the highest concentrations found along the Musi River corridor, decreasing outward from the riverbanks. Model validation achieved an AUC value of 0.968 (96.8%), demonstrating excellent predictive accuracy. These spatial outcomes provide actionable guidance for policymakers by identifying priority zones for sanitation and drainage upgrades, flood-resilient housing design, and targeted relocation planning. The study highlights the practical role of GIS-based quantitative modelling in supporting evidence-based slum management and urban infrastructure planning in Indonesia.