Relationships between Transportation Expenditures and Built Environment in the United States: Insights from Interpretable Machine-Learning Approach

Abstract

This study investigates the relationships between transportation expenditures and built environment (i.e., the D variables) in the United States using interpretable machine-learning (IML). Key findings include significant associations between these features, collectively contributing to 70.2 percent of feature importance. The study uncovers nonlinear relationships, such as transportation expenditures significantly declining after the compactness index value of 8. In addition, a pronounced reduction in transportation expenditures occurs when the job accessibility index falls below 4. This research contributes to (1) offering empirical evidence from nationwide analysis and (2) highlighting the potential of IML in improving our understanding, and (3) providing implications.