A novel approach for estimating multi-attribute Gaussian copula graphical models

Abstract

This paper considers estimating multi-attribute Gaussian copula graphical models where each node represents multivariate variables with rich meanings. A two-stage semiparametric method is proposed to achieve modeling flexibility and estimation robustness simultaneously by utilizing normal score transformation. We derive the consistency of the proposed estimator under the spectral norm, and establish the theoretical guarantees on sparsistency under some mild conditions. Simulation studies and a real data example are provided to demonstrate the empirical performance of the proposed method. We provide the complete code supporting this article at https://github.com/JerryLi-Stat/Multi-attribute-GCGM.