Applied Spatial Analysis and Policy Geography, Planning and Development
Identifying Urban Centers and Modeling Nonlinear Housing Price Gradients in a Polycentric Metropolitan Region: Geographically Weighted Regression and Tree-Based Approaches in the Seoul Metropolitan Area
Abstract
This study investigates the spatially nonlinear relationship between housing prices and urban centrality within the polycentric structure of the Seoul Metropolitan Area. Employing a hybrid analytical framework, we first identify multiple urban centers using kernel density estimation and spatial clustering techniques. Subsequently, we apply Geographically Weighted Regression (GWR) and Tree-Based approaches (DTR/RF) to model the complex, location-specific association between distance to the centers and housing prices. The results reveal significant spatial heterogeneity and nonlinear distance-decay patterns, indicating that proximity to different urban centers is associated with housing prices in varied and complex ways. These findings underscore the limitations of traditional monocentric assumptions and emphasize the importance of considering polycentric urban structures and spatial nonlinearity in urban planning and housing policy development.