What influences intermodal Choices: Metro-Centric, Bus-Centric, Hybrid? insights from Machine learning Approaches

Abstract

Three types of intermodals are defined: bus-centric, metro-centric, and hybrid, each representing combinations of bus, metro, and a mix of metro and bus with other travel modes for a trip, respectively. Using the household survey from Nanjing, China, comprising 162,191 trips, we applied the multiple models to reveal the nonlinear effects of socio-demographic and travel-related attributes on intermodal travel choices. Results show that bus-centric intermodal choice accounts for 65.82% of the total among the three types. The optimal model, random forest, indicates the relative importance of travel-related, individual, and household attributes, contributing 46.28%, 31.14%, and 22.59% respectively. Non-public transit travel time demonstrates an inverted V-shaped association with bus-centric intermodal choice, with a peak at around 5 min. Older individuals prefer bus-centric intermodal, while younger lean towards metro-centric and hybrid intermodal. Compared to car ownership and motorcycle ownership, bike ownership and E-bike ownership exhibit relatively high impact on intermodal travel choices.