A Multilayered Block Network Model to Forecast Large Dynamic Transportation Graphs, with an Application to US Air Transport

As network data become increasingly available, new opportunities arise to understand dynamic and multilayer network systems in many applied disciplines. Dynamic transportation networks have been analyzed for years by means of static graph-based indicators in order to study the temporal evolution of relevant network components, and to reveal complex dependencies that would not be easily detected by a direct inspection of the data. There is an opportunity for a methodological advance by using state-of-the-art statistical models for dynamic and multilayer graph data. Existing multilayer models are however typically limited to small, unstructured networks. In this paper we introduce a dynamic multilayer block network model with a latent space representation for blocks rather than nodes, which is natural for many real networks, such as social or transportation networks, where community structure naturally arise. The model and Bayesian inference are illustrated on a sample of 10-year data from the US air transportation system. We show how the proposed model projects the multilayer graph into the future for outof-sample full network forecasts, while stochastic blockmodeling allows for the identification of relevant communities and keeps estimation times within reasonable limits.