Modeling flight delays by an intensity‐based Hawkes process

Abstract

AbstractTwo key features of airline departure delays are that they cascade and that there can be exceptional peaks. We model these features using an intensity‐based Hawkes process. Our application to all KLM departure delays at Amsterdam Schiphol airport in January 2015 shows that volatility in departure delays is endogenous. We correlate the key parameters of the estimated Hawkes process with daily weather conditions and find that these conditions amplify the self‐exciting feature of departure delays.