Statistical Analysis of Aircraft Trajectories in a Multivariate Functional Data Analysis Framework
Résumé
While advanced inference and registration methods for functional data analysis have recently been developed in the literature, statistical analyses of aircraft trajectories have remained scarce, despite operational relevance. Using more than 3,000 trajectories matched with weather data, this paper explores how the experienced wind speed can be associated to en-route delays in a multivariate functional data analysis framework. The processing of aircraft trajectories is challenging as it requires constraint smoothing and rescaling. The paper emphasizes that the choice of the registration strategy influences further inference. Five scenarios are developed to compare registration strategies and find the most suited one for a pointwise functional two-sample test of means.
Origine : Fichiers produits par l'(les) auteur(s)