How weather and built environment factors influence e-scooter ridership: Understanding non-linear and time varying effects

Abstract

Our understanding of non-linear and time varying effects on shared e-scooter ridership dynamics is limited. Consequently, both operators and city councils supporting shared e-scooter schemes do not have the requisite information to help optimise infrastructure planning and operation management. Focussing on subtropical Brisbane, Australia, the current study examines time varying and non-linear effects of weather and built environment factors on shared e-scooter ridership. Results from XGBoost models reveal threshold relationships with both the availability of cycling infrastructure and the presence of park and commercial land uses. Additionally, we show how hot weather increases ridership especially around large parks and in commercial areas on both weekdays and weekends. Understanding the intricate (non-linear) interplay (interaction) between weather and built environment factors and their variation over time on shared e-scooter ridership have important implications for policymakers, transportation planners, and environmental advocates in providing the requisite evidence for data driven decision making.

Keywords

• Shared e-scooters
• Weather
• Built environment
• XGBoost
• Non-linear regression