Robust feature screening for elliptical copula regression model (Q2274965)

This paper presents the adaptive robust screening for the Elliptical Copula regression (ECR) model. The authors introduce the ECR model in the following way. Let \(f=\left\lbrace f_{0},f_{1},\ldots,f_{p} \right\rbrace \) be a set of monotone univariate functions. A random vector \(\mathbf{Z} = \left(Y,X_{1},\ldots, X_{p} \right)^{\mathrm{T}} \) satisfies the ECR model if and only if the vector \( \left( f_{0}\left(Y \right), f_{1}\left(X_{1} \right), \ldots, f_{p}\left(X_{p} \right) \right))^{\mathrm{T}} \) is elliptically distributed and \[ f_{0}\left(Y \right)= \sum_{j=1}^{p} \beta_{j} f_{j}\left(X_{j} \right) + \epsilon , \] where \(Y\) is the response, \(\left(X_{1},\ldots, X_{p} \right)^{\mathrm{T}}\) are covariates and \(\epsilon\) is the error term. The screening procedure is based on sorting canonical correlation estimated by Kendall's \(\tau\). From the authors' abstract: Theoretical analysis shows that the procedure enjoys sure screening property, i.e., with probability tending to 1, the feature screening procedure selects out all important variables and substantially reduces the dimensionality to a moderate size against the sample size. Thorough numerical studies are conducted to illustrate its advantage over existing feature screening methods. At last, the proposed procedure is applied to a gene-expression real data set to show its empirical usefulness.