Studying Understudied Populations’ Travel Behaviors with a Machine Learning Approach: A Focus on Hispanic and Latinx Households

To address a gap in the literature regarding differential effects of built environments (BE) on minority populations, this study uses boosted regression tree modeling to explore travel behaviors of Hispanic and Latinx (H/L) households, finding significant differences in the variables that influenced these households compared to others. Overall, BE variables had greater impacts on H/L households than others. Nonlinear effects (effective ranges, thresholds) were identified for top-ranking BE contributors: transit stop density, activity density, land use mix, and destination accessibility. Compact development is shown to have increased impacts on reducing vehicle miles traveled (VMT) for these populations.