A Better Way To Find The Best Flights And Avoid The Worst Airports

Abstract

US domestic aviation experienced approximately 80 million minutes of net delay across 6 million flights in 2014. Conventional performance metrics, which rely on binary on-time thresholds and self-reported delay causes, often fail to distinguish between airline operational efficiency and exogenous geographic constraints. To address these limitations, a regression-based methodology evaluates carrier performance by isolating airport-specific delays from actual flight durations. This model establishes a “target time” for every route, adjusted for distance and directional jet stream effects, to calculate the “time added” by specific airlines and airports. Analysis reveals that while major hubs such as Chicago O’Hare and the New York metropolitan airports contribute significantly to system-wide latency, substantial performance disparities persist among carriers even after adjusting for these factors. Highly delayed, canceled, or diverted flights—representing less than 5% of total operations—account for nearly 80% of total delay minutes. Furthermore, airlines employ varying degrees of schedule padding, which can mask underlying inefficiencies in standard government reporting. Findings indicate that carriers such as Virgin America and Delta maintain superior operational velocity, whereas United and American Airlines exhibit the highest adjusted delays. This multi-variable approach provides a more rigorous framework for assessing airline reliability by neutralizing the impact of route-specific variables and scheduling maneuvers. – AI-generated abstract.