Statistical inference based on robust low-rank data matrix approximation (Q2448728)

Based on the authors' abstract: The singular value decomposition is widely used to approximate data matrices with lower rank matrices. \textit{X. Feng} and \textit{X. He} [Ann. Appl. Stat. 3, No. 4, 1634--1654 (2009; Zbl 1184.62182)] developed tests on dimensionality of the mean structure of a data matrix based on the singular value decomposition. However, the first singular values and vectors can be driven by a small number of outlying measurements. In this paper a robust alternative that moderates the effect of outliers in low-rank approximations is considered. The asymptotic representations of the robust low-rank approximation is provided under the assumption of random row effects. These representations may be used in testing the adequacy of a low-rank approximation. Gene microarray data are used to demonstrate how robust singular value decomposition compares with its traditional counterparts.