The Minimum Regularized Covariance Determinant Estimator

27 Pages Posted: 3 Feb 2017 Last revised: 1 Dec 2018

See all articles by Kris Boudt

Kris Boudt

Ghent University; Vrije Universiteit Brussel; Vrije Universiteit Amsterdam

Peter Rousseeuw

KU Leuven - Department of Mathematics

Steven Vanduffel

Vrije Universiteit Brussel (VUB)

Tim Verdonck

KU Leuven

Date Written: December 1, 2018

Abstract

The Minimum Covariance Determinant (MCD) approach estimates the location and scatter matrix using the subset of given size with lowest sample covariance determinant. Its main drawback is that it cannot be applied when the dimension exceeds the subset size. We propose the Minimum Regularized Covariance Determinant (MRCD) approach, which differs from the MCD in that the { scatter} matrix is a convex combination of a target matrix and the sample covariance matrix { of the subset}. A data-driven procedure sets the weight of the target matrix, so that the regularization is only used when needed. The MRCD estimator is defined in any dimension, is well-conditioned by construction and preserves the good robustness properties of the MCD. We prove that so-called concentration steps can be performed to reduce the MRCD objective function, and we exploit this fact to construct a fast algorithm. We verify the accuracy and robustness of the MRCD estimator in a simulation study and illustrate its practical use for outlier detection and regression analysis on real-life high-dimensional data sets in chemistry and criminology.

Keywords: Breakdown value, High-dimensional data, Regularization, Robust covariance estimation

JEL Classification: C14, C32

Suggested Citation

Boudt, Kris and Rousseeuw, Peter and Vanduffel, Steven and Verdonck, Tim, The Minimum Regularized Covariance Determinant Estimator (December 1, 2018). Available at SSRN: https://ssrn.com/abstract=2905259 or http://dx.doi.org/10.2139/ssrn.2905259

Kris Boudt (Contact Author)

Ghent University ( email )

Sint-Pietersplein 5
Gent, 9000
Belgium

Vrije Universiteit Brussel ( email )

Pleinlaan 2
http://www.vub.ac.be/
Brussels, 1050
Belgium

Vrije Universiteit Amsterdam ( email )

De Boelelaan 1105
Amsterdam, ND North Holland 1081 HV
Netherlands

Peter Rousseeuw

KU Leuven - Department of Mathematics ( email )

Celestijnenlaan 200 B
Leuven, B-3001
Belgium

Steven Vanduffel

Vrije Universiteit Brussel (VUB) ( email )

Pleinlaan 2
Brussels, Brabant 1050
Belgium

HOME PAGE: http://www.stevenvanduffel.com

Tim Verdonck

KU Leuven ( email )

Celestijnenlaan 200B
Leuven, 3001
Belgium

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
135
Abstract Views
1,685
Rank
349,959
PlumX Metrics