There is growing recognition of the importance of cycling infrastructure in promoting cycling. However, established methods have a limited ability to predict the citywide effects of single infrastructure interventions on cycling. Network analysis provides a promising alternative; however, studies linking network measures to mode choice are rare. To address this gap, we applied two route-based network measures – route betweenness and the novel measure route link density – to evaluate the impact of the proposed bridge on cycling probabilities. Route betweenness measures the mean betweenness of shortest routes between origin–destination pairs, accounting for route connectivity, directness, and urban density. Route link density quantifies the mean density of network links along shortest routes. Using travel survey data, we trained a gradient boosting model to estimate cycling probabilities using route-based network measures and built environment variables at trip origins and destinations. We then applied the model to a synthetic population and estimated the changes in probability due to the bridge. The results show that route length and route-based measures are key features for predicting cycling probabilities. The models also reveal non-linear effects and notable feature interactions. When applied to a synthetic population, high route betweenness and moderate route link density increase the predicted probability of cycling. Conversely, low values can decrease probabilities, even when route lengths decrease. The effect of route betweenness is particularly strong, as the measure responds to the rerouting of trips over the bridge. Our findings highlight the importance of incorporating route-based measures in travel research and the need for longitudinal follow-up studies.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号