B1332
Title: Dynamic network models with time-varying nodes
Authors: Luca Gherardini - University of Florence (Italy) [presenting]
Abstract: Networks are used to represent complex data structures that arise in different fields of science and are often not static objects, as they can evolve over time. The main approaches in the literature belong to the broad class of latent variable models, including latent space models and stochastic block models. However, most methods were developed to model the dynamic behaviour of edges without considering that the network's topology may vary over time. It is expected that ignoring this new source of complexity can lead to distortions in the parameter estimates of the network model since the model is not able to distinguish between observed missing edges and the lack of edges for pairs of nodes that do not belong to the network topology at a given time. To address this relevant issue, a fully dynamic modelling framework is developed for undirected binary networks, which takes into account both the node and edge temporal behaviour. A class of zero-inflated Bernoulli models is proposed for the network edges, which discriminates between structural zeros for missing edges and those produced by observing a lack of edges between pairs of observed nodes. The inference approach for this class of models is developed within the Bayesian paradigm and relies on a Gibbs sampling algorithm with a Polya-Gamma data augmentation scheme. The performance of the approach is explored through a simulation study.