Toward better equity: Analyzing travel patterns through a neural network approach in mobility-as-a-service

The quantitative implications of Mobility-as-a-Service (MaaS) for travel equity and transportation system performance are underexplored, a gap primarily attributed to the complexities of forecasting travel pattern shifts in response to a broad spectrum of MaaS bundles via conventional methodologies. In this study, we utilize Neural Network models to identify the primary characteristics that influence the use of various travel modes, drawing upon data from a stated preference (SP) survey administered in Hong Kong. We evaluate the repercussions of enforcing various equity metric constraints on system efficiency and undertake a thorough analysis of the ensuing variations in system emissions and profitability. Our findings highlight the profound influence of established pre-MaaS travel patterns on subsequent behaviors following MaaS introduction. Moreover, our findings reveal that socio-demographic characteristics exert a greater impact on travel mode usage than the attributes of MaaS bundles, highlighting the entrenched lifestyle preferences and economic conditions that shape travel behavior. Furthermore, our research reveals that while the enforcement of equity constraints typically reduces system efficiency, specific ranges exist where such enforcement may be advantageous in terms of emission reductions and increased profitability for MaaS platforms.