Decoding travel behavioral intentions under flight delays via interpretable machine learning: Insights for safeguarding passenger mobility

Abstract

Understanding passenger behavior under flight delays is crucial for developing proactive policies that mitigate disruption-induced adverse effects. To support more effective and foresighted interventions, this study conducted a joint revealed preference and stated preference (RP-SP) survey at Beijing Daxing International Airport (BDIA) to analyze travel behavioral intentions in delayed trips. An Extreme Gradient Boosting (XGBoost) model was employed to elucidate the relationships between travel choice shifts and a set of explanatory variables, including socio-demographic attributes, travel characteristics, perceived service quality at the airport, and delay scenario features. The results show that socio-demographic attributes (e.g., work type, age) and travel characteristics (e.g., ticket price) hold higher relative importance in interpreting travel behavioral intentions. It is therefore necessary to implement differentiated service strategies tailored to passenger groups with different behavioral intentions. Also, findings reveal that the spatial variable matters in trip cancellation and highlight the importance of expanding high-speed railway as an alternative during flight disruptions in underserved regions. By identifying key determinants and ranking their importance in interpreting passenger behavior changes via machine learning instead of traditional econometric models, this study advances disruption management by offering a practical framework for user profiling-driven service strategies against flight delays. It further informs the airport/airline operators in optimizing resource allocation by implementing anticipatory and differentiated policy interventions towards higher operational resilience in preparation for future disruptions. The insights help ensure that delayed passengers can complete their trips successfully or make smooth adjustments to travel choices, supported by services that align with individual needs and ultimately enhance the overall travel experience.