Exploring non-linear and synergistic effects of green spaces on active travel using crowdsourced data and interpretable machine learning

The relationship between green spaces and active travel has been extensively studied. However, the majority of previous studies relied on small datasets concerning active travel and inadequately explored non-linear and/or synergistic effects. This study uses multi-source data and interpretable machine learning techniques to identify the non-linear and synergistic effects of green spaces in Chengdu (China) on two types of active travel: cycling and running. Crowdsourced data from Strava collected in December 2021 is used to measure city-wide active travel levels. Meanwhile, green spaces are evaluated from two viewpoints: overhead view and eye level, with the latter assessed using Baidu Street View imagery. The findings demonstrate that green spaces can account for up to 20% of the variance in active travel. Generally, the effect of the area of green spaces on active travel is positive. When the area of green spaces reaches a certain threshold, its effect becomes marginal and even negative. The green view index displays complex effects on cycling. Furthermore, this study identifies synergistic effects among predictors (e.g., green view index & river line length).

Automated summary

This study investigates the non-linear and synergistic effects of green spaces on active travel in Chengdu using multi-source data and interpretable machine learning techniques. The findings suggest that green spaces have a positive effect on active travel, but this effect diminishes and can even become negative when the area reaches a certain threshold. The green view index has complex effects on cycling, and there are synergistic effects among predictors.