Assessing the impacts of transit level-of-service and reliability on modal split

Implementing mobility services for low-income groups is crucial for promoting equity and sustainability in U.S. cities, where access to opportunities is often unequal. Planning a multi-modal system that meets the needs of marginalized populations while encouraging drivers to shift to transit services requires an understanding of the difference between user groups in the effects of transit service quality and reliability on model split. This study employs machine learning techniques to model the commuting mode choices of low-income and general users, using data for the Dallas/Fort Worth metropolitan area over 15 years. By analyzing built environment, transportation, socio-demographics, and transit level-of-service variables, we find that low-income user groups rely more on public transit when it offers greater coverage and accessibility. Meanwhile, the general user population tends to favor single-mode transit services with fewer transfers and shorter waiting times. Penalty time has a greater impact on low-income individuals, while built environment factors have a stronger effect on the general population. Our results provide insights into developing equitable, multimodal transportation systems.