Factors associated with discrepancies between scheduled and actual flight block times: evidence from U.S. domestic airline operations

Scheduled block time (SBT) represents the planned gate-to-gate duration of a flight and serves as a key element of airline schedule planning. In practice, actual block times (ABTs) frequently differ from scheduled values due to operational and environmental conditions. This study investigates factors associated with relative block time discrepancy between ABT and SBT using a dataset of approximately 7.13 million U.S. domestic flights in 2024, integrating flight operations data with airport characteristics, aircraft information, origin-airport weather observations, and airport traffic measures. The analysis combines descriptive exploration with multiple empirical modeling approaches, including robust linear regression, LASSO regression, multivariate adaptive regression splines (MARS), generalized additive models (GAM), Random Forest, and XGBoost. Results reveal systematic patterns in block time discrepancies across time of day, airport categories, and aircraft types. Weather and airport traffic variables remain important correlates, with precipitation, wind speed, humidity, and visibility showing relatively consistent associations across model specifications, while the role of temperature is more sensitive to fixed-effect controls. Regional jet operations show greater sensitivity to weather conditions, while congestion-related factors play a stronger role in wide-body operations. A common time-blocked out-of-sample comparison shows that all fitted models outperform a simple historical operational benchmark, with tree-based machine-learning models providing the strongest predictive performance. However, the incremental gains over other fitted models are modest, and overall explanatory power remains limited. Overall, the findings provide large-scale empirical evidence on the operational conditions associated with discrepancies between scheduled and actual block times and offer diagnostic insights for schedule reliability, congestion management, and operational performance.