ABSTRACT
In times of great uncertainty for the airline industry, travelers are in search of reliable itineraries now more than ever. With condensed airline schedules and less options, air travelers must rely on making flight connections and manage layover times to arrive at their final destination on time. In an era with readily available information, passengers expect accurate and transparent reliability information to help improve decision making for multi-leg itineraries. However, often for reliability in air travel, this information is incomplete or not useful. In this paper we utilize historical probability distributions of flight arrival and departure times using publicly available data to give an intuitive and predictive flight itinerary reliability metric. The COVID-19 pandemic significantly affected air-travel in the US and this uncertainty is still being felt with cancellations and delays due to staff shortages and reduced demand. Therefore, we extend the stochastic network model from our previous research to air travel during COVID-19 to see the effects on flight reliability. Using this model, we conduct computational experiments to evaluate air travel through multiple reliability metrics. We show that during periods of high uncertainty, predictive historical distributions of flight data considering recency and seasonal effects are less accurate given many cancellations and a reduced flight schedule. © 2022 Elsevier Ltd
ABSTRACT
In times of great uncertainty for the airline industry, travelers are in search of reliable itineraries now more than ever. With condensed airline schedules and less options, air travelers must rely on making flight connections and manage layover times to arrive at their final destination on time. In an era with readily available information, passengers expect accurate and transparent reliability information to help improve decision making for multi-leg itineraries. However, often for reliability in air travel, this information is incomplete or not useful. In this paper we utilize historical probability distributions of flight arrival and departure times using publicly available data to give an intuitive and predictive flight itinerary reliability metric. The COVID-19 pandemic significantly affected air-travel in the US and this uncertainty is still being felt with cancellations and delays due to staff shortages and reduced demand. Therefore, we extend the stochastic network model from our previous research to air travel during COVID-19 to see the effects on flight reliability. Using this model, we conduct computational experiments to evaluate air travel through multiple reliability metrics. We show that during periods of high uncertainty, predictive historical distributions of flight data considering recency and seasonal effects are less accurate given many cancellations and a reduced flight schedule.