Prediction of New Itinerary Markets for Airlines via Network Embedding

Dmitrii Kiselev, Ilya Makarov

Abstract: Network planning is one of the most significant problems that airlines solve every day. Currently, airlines utilise traveller decision choice modelling, which has certain drawbacks. It analyses each market independently, which does not consider the entire Airline network information with its dynamic structure formation based on competition factors. In the paper, we show that Airline network structure provides an accurate prediction for the current network and for future lines. We compare several approaches for Airline network link prediction via structural network embeddings, which are interpreted as new itinerary markets estimation.