Cross-Validated Loss-Based Covariance Matrix Estimator Selection in High Dimensions

TL;DR

This paper introduces a cross-validated loss-based method for selecting optimal covariance matrix estimators in high-dimensional settings, providing theoretical guarantees and practical tools for improved statistical analysis.

Contribution

It develops a general framework for estimator selection using cross-validation, with finite-sample risk bounds and asymptotic optimality results.

Findings

The proposed method outperforms traditional estimators in simulations.

Application to transcriptome data demonstrates practical effectiveness.

Provides an open-source R package for implementation.

Abstract

The covariance matrix plays a fundamental role in many modern exploratory and inferential statistical procedures, including dimensionality reduction, hypothesis testing, and regression. In low-dimensional regimes, where the number of observations far exceeds the number of variables, the optimality of the sample covariance matrix as an estimator of this parameter is well-established. High-dimensional regimes do not admit such a convenience, however. As such, a variety of estimators have been derived to overcome the shortcomings of the sample covariance matrix in these settings. Yet, the question of selecting an optimal estimator from among the plethora available remains largely unaddressed. Using the framework of cross-validated loss-based estimation, we develop the theoretical underpinnings of just such an estimator selection procedure. In particular, we propose a general class of loss functions for covariance matrix estimation and establish finite-sample risk bounds and conditions for the asymptotic optimality of the cross-validated estimator selector with respect to these loss functions. We evaluate our proposed approach via a comprehensive set of simulation experiments and demonstrate its practical benefits by application in the exploratory analysis of two single-cell transcriptome sequencing datasets. A free and open-source software implementation of the proposed methodology, the cvCovEst R package, is briefly introduced.