B0431
Title: A multiview network model for commodities trading data
Authors: Riccardo Rastelli - University College Dublin (Ireland) [presenting]
Chaonan Jiang - University of Pennsylvania (United States)
Davide La Vecchia - University of Geneva (Switzerland)
Abstract: A new class of latent space models is introduced to analyze the import/export trade data between a number of European countries. It is assumed that the probability of having a commercial relationship between two countries often depends on some unobservable (or not easy-to-measure) factors, like socioeconomic conditions, political views, and level of the infrastructure. To conduct inference on this type of data, a novel class of latent variable models is introduced for multiview networks, where a multivariate latent Gaussian variable determines the probabilistic behaviour of the edges. The model is labelled the graph generalized linear latent variable model (GGLLVM) and the inference is based on the maximization of the Laplace-approximated likelihood. The resulting M-estimator is called the graph Laplace-approximated maximum likelihood estimator (GLAMLE) and its statistical properties are studied. Using simulations and the real data application, the novel approach is demonstrated to be very computationally advantageous and that it can well capture many features of interest from the network.
