Key points are not available for this paper at this time.
Revealing the relationship between housing prices and Transit-Oriented Development (TOD) is important for housing market management and TOD planning. Research on TOD and housing prices has paid limited attention to simultaneously capturing the spatial heterogeneity and nonlinear effects of TOD features on prices. This study introduces a combined model – the optimized spatial proximity Geographically Neural Network Weighted with Extreme Gradient Boosting (osp-GNNW-XGBoost) model – which integrates the optimized spatial proximity Geographically Neural Network Weighted Regression (osp-GNNWR) and Extreme Gradient Boosting (XGBoost) to effectively evaluate the impact of Urban Transit Station Area (UTSA) features on housing prices in Nanjing, China. The walkability index, transportation index, proportion of park and cultural land use, and land use diversity are key UTSA features affecting prices, contributing 13.34%, 3.91%, 1.57%, and 1.55%. In addition, the study reveals the nonlinear and spatially heterogeneous mechanisms through which these key UTSA features affect housing prices. Furthermore, osp-GNNW-XGBoost shows higher prediction accuracy compared to XGBoost and osp-GNNWR, with an R² of 0.525, improving by 17.9% and 21.5%%, respectively. This study offers a new perspective on understanding the mechanisms between TOD and housing prices, providing insights for developing fair urban strategies.HighlightsReveals nonlinear and spatial effects of transit station areas on housing prices.Walkability, transit access, and land use mix are key contributors affecting prices.The spatially aware machine learning model shows improved predictive performance.Provides planning strategies for transit-oriented areas across different urban zones.
Liu et al. (Thu,) studied this question.