Stable estimation of a covariance matrix guided by nuclear norm penalties

DOI10.1016/J.CSDA.2014.06.018zbMATH Open1506.62043DBLPjournals/csda/ChiL14arXiv1305.3312OpenAlexW2081371310WikidataQ42544533 ScholiaQ42544533MaRDI QIDQ1623701FDOQ1623701

Authors: Eric C. Chi, Kenneth Lange 

Publication date: 23 November 2018

Published in: Computational Statistics and Data Analysis (Search for Journal in Brave)

Abstract: Estimation of covariance matrices or their inverses plays a central role in many statistical methods. For these methods to work reliably, estimated matrices must not only be invertible but also well-conditioned. In this paper we present an intuitive prior that shrinks the classic sample covariance estimator towards a stable target. We prove that our estimator is consistent and asymptotically efficient. Thus, it gracefully transitions towards the sample covariance matrix as the number of samples grows relative to the number of covariates. We also demonstrate the utility of our estimator in two standard situations -- discriminant analysis and EM clustering -- when the number of samples is dominated by or comparable to the number of covariates.

Full work available at URL: https://arxiv.org/abs/1305.3312

Recommendations

• Condition-number-regularized covariance estimation
• A well-conditioned and sparse estimation of covariance and inverse covariance matrices using a joint penalty
• A well-conditioned estimator for large-dimensional covariance matrices
• Covariance matrix selection and estimation via penalised normal likelihood
• Estimation of covariance matrices based on hierarchical inverse-Wishart priors

zbMATH Keywords

covariance estimationregularizationdiscriminant analysiscondition numberEM clustering

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Estimation in multivariate analysis (62H12)

Cites Work

• Regularized linear discriminant analysis and its application in microarrays
• Title not available (Why is that?)
• Finite mixture models
• A constrained formulation of maximum-likelihood estimation for normal mixture distributions
• Bayesian regularization for normal mixture estimation and model-based clustering
• A well-conditioned estimator for large-dimensional covariance matrices
• Nonlinear shrinkage estimation of large-dimensional covariance matrices
• Sparse inverse covariance estimation with the graphical lasso
• Mathematical and statistical methods for genetic analysis.
• Regularized estimation of large covariance matrices
• Model selection through sparse maximum likelihood estimation for multivariate Gaussian or binary data
• Penalized Normal Likelihood and Ridge Regularization of Correlation and Covariance Matrices
• Partial correlation estimation by joint sparse regression models
• Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions
• Constrained monotone EM algorithms for finite mixture of multivariate Gaussians
• High-dimensional covariance estimation by minimizing \(\ell _{1}\)-penalized log-determinant divergence
• A likelihood-based constrained algorithm for multivariate normal mixture models
• Title not available (Why is that?)
• Estimation of high-dimensional low-rank matrices
• The variational form of certain Bayes estimators
• Wishart distributions for decomposable covariance graph models
• Nonparametric estimation of large covariance matrices of longitudinal data
• Title not available (Why is that?)
• Flexible covariance estimation in graphical Gaussian models
• Estimation of a covariance matrix under Stein's loss
• Minimax estimation of large covariance matrices under \(\ell_1\)-norm
• Randomized Algorithms for Matrices and Data
• Covariance matrix selection and estimation via penalised normal likelihood
• Covariance estimation: the GLM and regularization perspectives
• Shrinkage Estimators for Covariance Matrices
• A trace inequality of John von Neumann
• Hub Discovery in Partial Correlation Graphs
• Large-Scale Correlation Screening
• High-dimensional covariance matrix estimation in approximate factor models
• High-dimensional covariance estimation
• Simultaneous modelling of the Cholesky decomposition of several covariance matrices
• Estimation of the multivariate normal covariance matrix under some restrictions

Cited In (3)

• Estimating large correlation matrices for international migration
• An efficient numerical method for condition number constrained covariance matrix approximation
• The spectral condition number plot for regularization parameter evaluation

Uses Software

• UCI-ml
• rda
• glasso
• CondReg

This page was built for publication: Stable estimation of a covariance matrix guided by nuclear norm penalties

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1623701)