How the Built Environment Shapes Micromobility Activity? An Unsupervised Machine Learning Approach Using Boston Bluebikes Data
Author(s)
Li, Sitao
Advisor(s)
Akar, Gulsah
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
This study examines how built environment characteristics shape shared e-scooter trip patterns in an urban context. Trip-level temporal features and land-use indicators were compiled and reduced using Principal Component Analysis to capture key dimensions of activity intensity, temporal preference, and neighborhood functions. K-means clustering was then applied to identify groups of origins with distinct trip-generation behaviors. Results reveal clear spatial and temporal heterogeneity: some clusters are associated with mixed-use and commercial areas exhibiting strong peak-hour demand, while others reflect primarily residential contexts with more stable all-day activity. The findings highlight how
micromobility usage varies across neighborhoods and demonstrate the importance of integrating land use, temporal structure, and local urban form when planning for shared mobility systems. Despite data limitations, including incomplete land-use accuracy and the absence of demographic variables, the study offers an empirical framework. This framework can support more context-sensitive micromobility deployment strategies.
Sponsor
Date
2025-12-05
Extent
Resource Type
Text
Resource Subtype
Rights Statement
Unless otherwise noted, all materials are protected under U.S. Copyright Law and all rights are reserved