On dimension reduction in conditional dependence models

Abstract: Inference of the conditional dependence structure is challenging when many covariates are present. In numerous applications, only a low-dimensional projection of the covariates influences the conditional distribution. The smallest subspace that captures this effect is called the central subspace in the literature. We show that inference of the central subspace of a vector random variable $\mathbf Y$ conditioned on a vector of covariates $\mathbf X$ can be separated into inference of the marginal central subspaces of the components of $\mathbf Y$ conditioned on $\mathbf X$ and on the copula central subspace, that we define in this paper. Further discussion addresses sufficient dimension reduction subspaces for conditional association measures. An adaptive nonparametric method is introduced for estimating the central dependence subspaces, achieving parametric convergence rates under mild conditions. Simulation studies illustrate the practical performance of the proposed approach.