The association of subjective physical disorder and pedestrian volume: A big urban data and machine-learning approach

Physical disorder in an urban area is characterized by visible damage, decay, and deterioration in its built environment, such as broken windows, graffiti, and litter. While its adverse effects on mental health, crime rates, and life satisfaction are well-documented, its impact on pedestrian volume–an essential indicator of urban vibrancy and livability–remains poorly discussed. Moreover, previous studies have predominantly relied on objective measures of physical disorder, overlooking subjective perceptions and potentially leading to biased interpretations. To address these crucial research gaps, we developed an online visual survey to evaluate the perceived physical disorder in Shanghai, China, across five dimensions: architectural disorder, commercial disorder, road disorder, greenery disorder, and infrastructure disorder. Then, we leveraged diverse machine learning algorithms to predict citywide spatial patterns of physical disorder based on both high-level street elements and low-level features. Finally, we examined the associations between urban physical disorder and pedestrian volumes, categorized by age and gender. Our findings reveal disparities in the influence of different types of subjective physical disorder on pedestrian volumes by demographic groups. Moreover, the subjective physical disorder provides a valuable supplement to existing built environment factors in explaining collective walking behavior. Notably, greenery disorder exhibits a significant negative association with walking behavior among female, adult, and elderly pedestrians, whereas infrastructure disorder predominantly impacts young pedestrians. Leveraging big data, this subjective measurement framework enables demographically sensitive evaluation systems of physical disorder as well as targeted interventions to reduce perceived physical disorder and improve walkability for different population groups.