Analysis of the Spatial Heterogeneity of Nighttime Taxi Demand from the Perspective of Built Environment

In the field of urban transportation, exploring how the built environment influences the spatiotemporal variation of taxi travel demand is a critical issue. Based on this, this study integrates taxi GPS data, point of interest (POI) data, and road network data from Shenzhen, and constructs a built environment feature system from five dimensions: density, diversity, design, destination accessibility, and proximity to public transportation. The study adopts the Geographically Weighted Random Forest (GWRF) model, which integrates spatial weighting and localized random forests to capture spatial heterogeneity and nonlinear effects. Evaluation metrics show that GWRF (Rš=0.96, RMSE=120.358, MAE=46.731) outperforms traditional OLS and GWR models. The findings reveal the complex influence mechanisms of different built environment features on taxi travel demand, providing a scientific basis for optimizing urban transportation planning and management, especially with important application value in improving the quality and efficiency of transport services during periods of active nighttime economy.