Date/Time
Date(s) - September 27, 2019
11:45 am - 12:35 pm
Location
FLG 0245
Categories
Xilei Zhao, Ph.D.
Assistant Professor
Department of Civil & Coastal Engineering, University of Florida
Abstract: Data Science Applications in Transportation
This talk introduces two data science applications in transportation. The first application is travel behavior modeling using interpretable machine learning. I discuss the key differences between machine learning and logit models in modeling travel mode choice, focusing on model development, evaluation, and interpretation. I then apply various machine-learning interpretation tools (e.g., sensitivity analysis, gradient-based methods, and model distillation) and invent new ones to examine behavioral heterogeneity. The second application is an integrated model for transportation networks and travel time reliability. The transportation network is modeled as a two-level system, with links as subsystems and routes as full systems, using maximum likelihood estimation. I then formulate the Fisher Information Matrix and apply the asymptotic normality to obtain the probability distribution of travel time estimates for a route. The travel time reliability is thus modeled by considering two-level uncertainties, i.e., uncertainty from travel times and uncertainty from travel time estimates.