Causal and Spatial Safety Modeling

This project line focuses on the mechanisms behind crash risk rather than prediction alone. It uses causal mediation, double machine learning, causal forests, Bayesian hierarchical models, random-parameter models, and spatial effects to explain how roadway, traffic, visual, and behavioral factors shape safety outcomes.

Representative publications include work in Transportation Research Part C, Analytic Methods in Accident Research, Accident Analysis and Prevention, and IEEE Transactions on Intelligent Transportation Systems.