Title: Levy copulas in Bayesian non-parametric models
Authors: Alan Riva Palacio - University of Kent (Mexico) [presenting]
Fabrizio Leisen - University of Kent (United Kingdom)
Abstract: Completely random measures have been used to construct a variety of models in Bayesian non-parametric statistics. Generalization of such models into a multivariate setting can be done in the framework of Levy copulas. We highlight the family of Levy copulas defined by certain compound random measures. In particular we showcase a generalization of the Clayton Levy copula. We present an application to vectors of survival functions where inference is done in a fully Bayesian way that takes advantage of simulation algorithms associated to Levy copulas.