Menu Close

Complex Systems and Data Sciences: Towards a new Perspective for the Understanding of Air Transport


Seddik Belkoura


Massimiliano Zanin, Antonio Latorre de la Fuente

Research area

Big Data

Affiliated Research Center

Universidad Politécnica de Madrid

Complex systems, i.e. systems composed of a large set of elements transporting and interchanging information in a non-linear way, are constantly found all around us. In the last decades, the approach toward their understanding has shifted progressively from a transportation to an information processing point of view. In other words, we are moving from a movement-based analysis (i.e. tracking the movement of items through time and space to reconstruct various metrics about their behaviour) to a higher-level approach, where individual movements are left aside to focus on the distribution, processing and flow of the information within the system. The information processing approach presents the main advantage of being data-based, that is, that no a priori knowledge about the interactions in the system is needed, hence the absence of costly simulations models. Such paradigm perfectly fits within the air transport system, where thematics as important as delay propagation (for its economical, environmental and safety related consequences) has been until now mainly analysed from a transportation micro-level perspective. Yet, the progressive rise in aviation of data analyses encourages a more data-centred path. We here present the first work that aims at fostering the combined use of the intuitive microscopic point of view with a higher-level information processing approach, yielding a more complete characterisation of the delay propagation process. Specifically, the here propose a three-fold approach. First, we highlight the degree of subjectivity associated with network-based representations of the air transport system, which conditions the intelligence extracted from any information processing study. Secondly, we manufactur a new data mining technique to extract non-linear causality relationships, therefore enabling the creation of a more complete delay propagation network representation. Finally, we complement our results by a micro-level analysis, therefore ending up with a 360â—¦ view of the delay propagation process. These analysis have been performed mainly on a European dataset, but expanded to other airspaces whenever data have been available.