Identifying Nighttime Vitality Through Multisource Geodata: An Explainable AI Perspective

ABSTRACT Nighttime vitality serves as a crucial criterion for assessing the attractiveness and competitiveness of urban areas. The varying and multifaceted characteristics of nighttime vitality across neighborhoods reflect residents' preferences, social demands, and localized economies. The objective of this study is to comprehensively assess nighttime vitality by integrating multiple geospatial data sources, including mobile phone data, nighttime remote sensing data, and small catering businesses data. Furthermore, within an explainable artificial intelligence (XAI) framework, this study aims to quantitatively analyze the nonlinear relationships between the influencing factors and nighttime vitality. To enhance interpretability, we employ SHapley Additive exPlanations (SHAP) for visual model interpretation. This study can further enhance the theory of urban vitality, deepen the understanding of nonlinear relationships, and facilitate the implementation of explainable AI in urban studies. The findings of this study can effectively provide guidance for urban planning and decision‐making, improve residents' quality of life, and promote sustainable urban development.