A machine learning approach to rank pre-season flight schedules

Abstract

The rate of air traffic demand increase in Europe is exceeding the pace of capacity generating concept and infrastructure development, resulting in imbalances between traffic demand and capacity.To mitigate airport congestion, in the strategic phase, i.e., up to 6 months prior to the day of the flight execution, pre-season flight schedules are implemented at the busiest airports in Europe. The aim of the pre-season flight schedules is to limit the number of scheduled flights during peak hours.This paper proposes a machine learning approach to classify the pre-season scheduled flights as delayed or cancelled with high accuracy. We also identify the most significant features for the output of the classification algorithms. Further, we propose a generic method to rank pre-season flight schedules using a set of predefined, airport performance indicators. We employ this method to rank 10 pre-season flight schedules, where the performance indicators are derived from the flight delay and cancellation predictions. We apply the flight classification and ranking algorithms at London Heathrow Airport, one of the busiest airports in Europe. Together with the development of dedicated pre-season flight schedule optimization models, our proposed approach supports an integrated pre-season flight schedule assessment, where pre-season flight schedules are re-evaluated with respect to on-time airport performance.