City-scale unmet bus demand modeling with passenger reneging behavior

Abstract

Achieving a balance between transit supply and passenger demand is essential for ensuring reliable and equitable public transportation services. In high-demand urban bus systems, however, capacity constraints frequently lead to boarding failures, where passengers are unable to board and must either wait or abandon their trips. These events represent not only operational inefficiencies but also critical accessibility issues that disproportionately affect transit-dependent users. Despite its importance, unmet demand remains difficult to quantify because it is inherently unobserved in standard transit data. Existing approaches either rely on costly sensing infrastructure or focus on post-hoc analysis, limiting their applicability for city-scale and timely decision-making. Moreover, passenger behavioral responses—such as waiting tolerance and reneging—are rarely incorporated, leading to incomplete representations of demand dynamics. This study proposes a behaviorally grounded model to estimate latent unmet demand using smart card data. The approach explicitly captures passenger reneging as a function of expected waiting time, enabling a more realistic representation of demand propagation under capacity constraints. Using comprehensive smart card data from the Seoul metropolitan bus network, the proposed method improves estimation accuracy and enables real-time, city-scale monitoring. The findings reveal that unmet demand is highly dynamic and spatially dispersed, and that ignoring passenger behavior leads to systematic misestimation of service gaps. The proposed model provides actionable insights for transit operators and policymakers, supporting demand-responsive interventions to improve service reliability and accessibility.