<link rel="stylesheet" href="styles.cf390c6bfbe7555f.css">

Geospatial Perspectives on Human-Centered Transportation Using Big Mobile Device Data and Machine Learning Methods

Loading...
Thumbnail Image

Files

Zhao_umd_0117E_26230.pdf (4.45 MB) (RESTRICTED ACCESS)
No. of downloads:

Publication or External Link

External Link to Data Files

Date

Advisor

Stewart, Kathleen

Citation

Abstract

This dissertation examines how different population groups travel, perceive, and interact with transportation systems through three interconnected studies in Maryland, leveraging big mobile device location data, social media data, and machine learning methods. The first study used explainable machine learning methods with mobile device location data to examine travel behavior differences across population groups and identify factors associated with variations in mobility patterns. The second study developed a prompt-based large language model (LLM) framework to classify Google Maps reviews of transportation-related points of interest (POIs) and analyze spatial and socioeconomic patterns of public sentiment toward transportation facilities and services. The third study integrated public sentiment derived from social media with visit patterns estimated from mobile device location data and employed structural equation modeling to examine the associations between public sentiment, socioeconomic status, and transportation POI visit patterns. This dissertation contributes to both geospatial science and transportation planning by offering methodological frameworks that incorporate explainable machine learning, prompt-based LLMs, and diverse sources of location data to study human-environment interactions relevant to human mobility and urban analytics research, while providing transportation planners with information on how different population groups travel, how the public perceives transportation services, and how these perceptions are associated with facility visit patterns, collectively supporting the development of more human-centered transportation systems by bridging advances in geospatial data science with the practical needs of transportation planning and policy.

Notes

Rights

Endorsement

Review

Supplemented By

Referenced By