Modeling e-scooter route choices: Infrastructural preferences and intervention scenario analysis

The e-scooters offer a flexible and sustainable alternative to traditional transport modes. To maximize its benefits, cities need strategic infrastructure enhancements. However, there is still a lack of research on how built environment factors affect e-scooter routing in a nonlinear manner, and how the nonlinearity may affect policy interventions. To bridge this gap, this study draws inspiration from three-factor theory to classify built environment attributes into eight types and analyze how bike lanes interact with these attributes to influence route preferences. By employing a perturbed utility route choice model, we examine infrastructural preferences of e-scooter routing from extensive GPS trajectories. Our results show that roads with high tree coverage and asphalt bike lanes see a generalized cost reduction of 28.6% compared to roads with moderate tree coverage and no asphalt bike lanes, while roads with low tree coverage increase 16.2% of generalized cost relative to the same reference road type. This highlights an asymmetric and nonlinear influence of built environment attributes on e-scooter route choices. Scenario simulation suggests that increasing tree coverage from moderate to high on roads with asphalt bike lanes or constructing new asphalt bike lanes on roads with high tree coverage are particularly effective in reducing generalized route costs. Our study offers a more nuanced understanding of infrastructural preferences in e-scooter route choices, and provides an analytic framework to enhance e-scooter adoption and inform policy-making to cater to future micro-mobility needs.