Assessing Preference Heterogeneity for Mobility-on-Demand Transit Service in Low-Income Communities: A Latent Segmentation Based Decision Tree Method

Abstract

The future of public transit service is often envisioned as Mobility-on-Demand (MOD), i.e., a system that integrates fixed routes and shared on-demand shuttles. The MOD transit system has the potential to provide better transit service with higher efficiency and coverage. However, little research has focused on understanding traveler preferences for MOD transit and preference heterogeneity, especially among the disadvantaged population. This study addresses this gap by proposing a two-step method, called latent segmentation based decision tree (LSDT). This method first uses a latent class cluster analysis (LCCA) that extracts traveler profiles who have similar usage patterns for shared modes. Then, decision trees (DT) are adopted to reveal the associations between various factors with preferences for MOD transit across different clusters. We collected stated-preference data among two low-resource communities, i.e., Detroit and Ypsilanti, Michigan. The LCCA model divides the entire sample into three clusters, i.e., shared-mode users, shared-mode non-users, and transit-only users. We find that job accessibility by transit is the most important variable for all the cluster-specific DT to model the MOD transit preference, and it negatively associated with the MOD transit preference. For transit-only users, gender and car ownership are the second-important variables, but neither of them appears in the DT for the other two clusters. In particular, female transit-only users have lower preference for MOD transit, possibly due to safety concerns. The LSDT method can generate richer insights than a single DT fitted to the overall sample by better accounting for heterogeneity. The findings gained from this approach can inform better-targeted strategies to plan for MOD transit services.