Key points are not available for this paper at this time.
Understanding the intricacies of fine-grained population distribution, including both predictability and uncertainty, is crucial for urban planning, social equity, and environmental sustainability. The spatial processes associated with the distribution of populations are complex, and enhancing their predictability involves revealing nonlinear interactions among various explanatory variables. Additionally, population distribution is influenced by various factors that are often challenging to quantify, thereby introducing uncertainty into predictive models. Although the development of explainable artificial intelligence (XAI) helps identify underlying factors, the complex geographical processes and the special nature of spatial data present challenges for purely statistical-based explanation methods, leading to incomplete or incorrect explanations. To address these challenges, we introduce GeoVisX, a geospatial visual analytics framework integrated with XAI. GeoVisX integrates XAI with visual analytics to dissect the spatial processes. Through a case study of Munich, GeoVisX demonstrates its utility in analyzing spatial distribution and identifying key factors impacting population distribution at the 100 m grid level. Our findings highlight the GeoVisX’s capability to enhance understanding of geographical phenomena, contributing to more informed urban policy and planning strategies. This study not only validates the effectiveness of GeoVisX but also emphasizes the importance of incorporating visual analytics and explainable methodologies for addressing complex geographical issues.
Luo et al. (Tue,) studied this question.