Citation

Tuli FM, Mitra S. Travel Behav. Soc. 2024; 36: e100812.

Copyright

(Copyright © 2024, Elsevier Publishing)

DOI

10.1016/j.tbs.2024.100812

PMID

unavailable

Abstract

The study analyzed the survey data from the 2018 Portland E-scooter Pilot Program and aims to determine (i) who uses shared e-scooters and why they use them, and (ii) whether there is any association between e-scooter usage and the usage of other modes of transportation. To accomplish the first objective, the study identifies the users of shared e-scooters based on their travel behavior using an unsupervised machine learning approach, latent class analysis (LCA). The LCA model grouped e-scooter users into three distinct classes: Class 1 (Recreational Enthusiasts) −occasional and frequent users for recreation, Class 2 (Commute Riders) −frequent users for work, and Class 3 (Intermittent Joyriders) −occasional and one-time users for recreation. Furthermore, a set of ordered logit models is employed to determine the second objective based on the identified classes of e-scooter users, their socio-demographic characteristics, and the built environment variables. The results of ordered logit models revealed that compared to Commute Riders, both Recreational Enthusiasts and Intermittent Joyriders exhibit less interest in increasing the usage of available transportation modes after adopting e-scooters. Notably, low-income e-scooter users show a higher probability of increasing their usage across various transportation modes, including public transportation, driving, shared mobility services, personal bikes, shared bikes, and walking. The study offers valuable insights to guide city planners and policymakers in developing effective strategies for the deployment of e-scooters, targeting each group of users.

Keywords

E-scooters; Latent Class Analysis; Ordered Logit Model; Shared micromobility; User Segmentation