Mixture Copula Model On Correlated Parameter And Margin

In recent years, copula as a multivariate modeling method showed strong flexibility and ability, and has grown extremely fast. Abe Sklar(1959) initiatively incorporated cop-ula into Statistics and proved the famous Sklar theorem. Schweizer, Wolff(1981) discussed copula’s invariance which avoided the rigorous research on marginal distribution. It made feasible to construct copula model just base on the rank of the marginal samples. Using assumptions of Archimedean copula, Genest, Rivest(1993) developed a Kendall τ momen-t estimation. Oakes(1994), Genest, Ghoudi, Rivest(1995) developed pseudo-likelihood semi-parametric estimation.In order to enhance the model flexibility, Johnson et al(1993) discussed parameter-mixture copula. Chen(2006) also gave a method using partition sample in copula mod-eling. Inspired, The thesis incorporate Bayesian paradigm and assume parameter and marginal random variable distributed to a bivariate distribution, then construct a mix-ture Copula on correlated parameter and margin.