Journal of the Korean Cartographic Association ISSN:1598-6160(Print) 한국지도학회지

House prices are closely related to their locations, and their determinants also involve spatial heterogeneity, which refers to the variations in their effects between global and local models. Recently, spatial heterogeneity has been expanded to differentiate between continuous and discontinuous spatial heterogeneity based on patterns of spatial variation. This study explores the spatial heterogeneity of apartment sales price determinants in Seoul from both continuous and discontinuous perspectives. Specifically, the study implements both geographically weighted regression (GWR) and regionally geographically weighted regression (RGWR) models, with the latter incorporating regional constraints into the GWR. The study mainly examines the spatial variation of local regression coefficients both within and between regions, and further compares the two models based on prediction accuracy and the spatial autocorrelation of residuals. The analysis results show that determinants such as the year of construction highlight the discontinuities between regions in the RGWR without obscuring the spatial variations within a region, providing better explanations from the perspective of discontinuous spatial heterogeneity. The study also confirms that the perspective of continuous spatial heterogeneity remains valid for determinants such as the neighborhood park distance, which maintain continuity both between and within regions. Finally, the RGWR model empirically demonstrates improved results over the GWR model, with higher prediction accuracy and lower positive spatial autocorrelation in the residuals. This study is expected to extend the understanding of spatial heterogeneity, particularly in housing price determinants, and serve as a foundational study for categorizing spatial heterogeneity.