Revealing land-use functional patterns and associated factors in China’s high-speed rail station areas: Based on self-organizing maps and explainable machine learning algorithms

The rapid expansion of high-speed rail (HSR) is closely associated with urban spatial restructuring. However, HSR station areas show substantial differentiation in land-use structure and functional positioning, and their functional patterns and associated characteristics remain insufficiently understood at the national scale. Taking 1570 HSR station areas in China as the research objects, this study constructs an integrated framework of comparative advantage, SOM clustering, and XGBoost-SHAP to identify land-use functional patterns in HSR station areas and analyze their associated factors. The results show that HSR station areas can be classified into four typical patterns: Public facility-dominated pattern (PFDP), Balanced multifunctional pattern (BMP), Industry-dominated pattern (IDP), and Commercial service-dominated pattern (CSDP). Their spatial distribution is closely related to regional industrial bases, city hierarchy, and the structure of the HSR network. The XGBoost-SHAP analysis further indicates that land-use diversity, bus accessibility, road network length, industrial structure, and distance to the city center are key variables closely associated with functional differentiation in station areas, and they show significant nonlinear responses and interaction-related characteristics in the model predictions for different functional patterns.This study contributes to a deeper understanding of functional differentiation in HSR station areas and provides theoretical support and planning guidance for HSR new town development, land use optimization, and urban functional restructuring.