Research on the impact of urban climate factors on planning and design based on the XGBoost regression model

The impact of urban climate factors on planning and design has attracted increasing attention, particularly in the context of extreme weather events and global warming. Given its geographic diversity and climatic variability, the United States— represented here by 49 states—offers an ideal dataset for large-scale modeling. This study employs the XGBoost (Extreme Gradient Boosting) regression model to analyze multiple meteorological indicators, including temperature, humidity, wind speed, and visibility. SHAP (SHapley Additive exPlanations) is further applied to interpret the decision process of the model and to explore the interaction mechanisms among key features. Results indicate that temperature plays a dominant role in the model’s predictions, while humidity, wind speed, and visibility exhibit moderate to strong influence. Notably, a significant interaction effect between temperature and humidity is observed, wherein high-temperature and high-humidity conditions jointly amplify the model’s output. The XGBoost model demonstrates distinct advantages in capturing complex interactions and improving prediction performance. This study provides a feasible approach for broader climate– urban coupling research and highlights the potential of combining machine learning with interpretable analytics in supporting scientific decision-making in urban planning.