Explaining temporal and spatial variability in shared e-scooter ridership in Christchurch, New Zealand

Shared electric scooters (e-scooters) are widely adopted, yet their ridership patterns remain insufficiently understood. Existing studies rely on origin–destination (OD) inferred trips and use global models, which overlook true paths and spatial heterogeneity effects. This constrains understanding of spatio-temporal patterns and drivers of shared e-scooter ridership, reducing the effectiveness of transport planning and policy interventions. We address these limitations using observed street-segment-level trip counts to identify spatio-temporal ridership profile by clustering hourly ridership curves. We also apply Geographically Weighted Poisson Regression (GWPR) to examine spatial associations between ridership and contextual factors. Results reveal clear temporal differences from cycling behaviour and strong spatially varying associations between ridership and built environment, which global models fail to capture. Our findings show that data-driven clustering can classify micromobility behaviour without assuming trip purpose. Although focused on Christchurch, the methodological framework is transferable and applicable to other cities with similar data.