Discovering impacts of built environment on transit ridership in the post-COVID era:policy intervention and nonlinear dynamicsOA

Transit ridership continues to struggle with recovery in the post-COVID era.While previous studies have examined ridership patterns after COVID-19,the collective impact of lifted control policies and built environment factors on ridership recovery remains unclear.This study merges regression discontinuity design(RDD)with random forest(RF),and proposes a hybrid RDD-RF model,to estimate the effects of COVID-related policies,temporal dynamics,and built environment factors on ridership.The model leverages RDD''s causal inference(CI)and RF''s flexibility to handle complex interactions,enhancing predictive accuracy and interpretability.An empirical analysis in Nanjing,China,using transit data from 2020,2021,and 2023,shows that reopening policies significantly impact ridership,but rank fourth behind bus routes,station entry/exit points,and restaurants.Commuting and daily life needs are key drivers of recovery.We also identify effective ranges and thresholds for built environment variables,providing targeted strategies to boost transit ridership in the post-COVID era.