Nonparametric Covariate Adjustment in Copula-based Dependence Models
by Dr Elif Acar
Abstract: Adjusting statistical dependence for covariates via conditional copulas is an active area
of research where model ﬁtting and validation are currently in early development. This talk presents a uniﬁed framework to assess and infer covariate effects on the dependence structure of random variables in bivariate or multivariate models. In conditional copulas, the copula parameter is deterministically linked to a covariate via the calibration function. To estimate the latter, we propose a nonparametric estimation procedure based on local likelihood and use the proposed estimator as a diagnostic tool for testing a parametric formulation of the calibration function against a general alternative. We demonstrate the estimation and test procedures using subsets of the Matched Multiple Birth and Framingham Heart Study datasets. Multivariate extensions via pair-copula constructions are also addressed.
For More Information: for further information contact Prof Richard Huggins. email: email@example.com