Proposal of a Model Selection and Parameter Estimation for D-vine t Copulas Constructions Model in Discrete Data

Abstract

Abstract. In the case of discrete data, we propose a novel method for both model selection and parameter estimation by detecting independence copulas from a model whose dependence structure is constructed by a Dvine t copula using the Reversible Jump Markov Chain Monte Carlo together with the Data Augmentation. It was shown that the proposed method has sufficient parameter estimation accuracy.