A0505
Title: Conditional copula models using loss-based Bayesian additive regression trees
Authors: Tathagata Basu - Newcastle University (United Kingdom) [presenting]
Fabrizio Leisen - Kings College London (United Kingdom)
Cristiano Villa - Duke Kunshan University (China)
Kevin Wilson - Newcastle University (United Kingdom)
Abstract: The aim is to present a novel semi-parametric Bayesian approach for modeling conditional copulas to understand the dependence structure between two random variables when it is influenced by a different covariate. The use of Bayesian additive regression trees is proposed to model the conditional copulas. A loss-based prior is specified for the BART model suggested by a prior study, which is designed to reduce the loss in information and complexity for tree misspecification, giving a parsimonious model that avoids over-fitting, a common issue of BART models. Results are presented with both simulated and a real dataset to show the applicability and efficiency of the method.
