Speaker
Description
Urban green spaces (UGS) enhance resident well-being, yet effective planning requires understanding the complex, often nonlinear relationships that shape park use, as subtle changes in perceived environments can significantly impact engagement. However, most research relies on linear models, constraining the exploration of these nuanced interactions, and primarily focuses on park-specific or demographic factors, overlooking the broader residential context’s role in shaping visitation patterns. This study addresses these gaps by integrating street view imagery (SVI) and mobile data to examine nonlinear relationships between perceived residential environments and park visitation across Tokyo's 23 special wards using Gradient Boosted Decision Trees (GBDT). Results indicate over 41% of the feature importance attributed to perceptions of park visitation frequency and diversity. 'Depressing' environments were linked to lower visitation, while 'beautiful' and 'wealthy' environments showed threshold effects, with notable reversals as perceptions increased from mid to high. 'Safety', however, was relatively less influential. These insights indicate that park planning could be enhanced by targeted strategies, such as fostering community engagement in 'depressing' areas and introducing distinctive amenities in 'wealthy' neighborhoods, to better align with local needs and encourage consistent park usage.
| Keywords | Urban green spaces; Perceived residential environment; Human mobility |
|---|---|
| Best Congress Paper Award | Yes |
