The distribution of robust distances

Johanna Hardin, David M Rocke

Research output: Contribution to journalArticle

104 Citations (Scopus)

Abstract

Mahalanobis-type distances in which the shape matrix is derived from a consistent, high-breakdown robust multivariate location and scale estimator have an asymptotic chisquared distribution as is the case with those derived from the ordinary covariance matrix. For example, Rousseeuw's minimum covariance determinant (MCD) is a robust estimator with a high breakdown. However, even in quite large samples, the chi-squared approximation to the distances of the sample data from the MCD center with respect to the MCD shape is poor. We provide an improved F approximation that gives accurate outlier rejection points for various sample sizes.

Original languageEnglish (US)
Pages (from-to)928-946
Number of pages19
JournalJournal of Computational and Graphical Statistics
Volume14
Issue number4
DOIs
StatePublished - Dec 2005

Fingerprint

Minimum Covariance Determinant
Covariance matrix
Breakdown
Chi-squared
Robust Estimators
Approximation
Rejection
Asymptotic distribution
Outlier
Sample Size
Estimator

Keywords

  • Mahalanobis squared distance
  • Minimum covariance determinant
  • Outlier detection
  • Robust estimation

ASJC Scopus subject areas

  • Mathematics(all)
  • Statistics and Probability
  • Computational Mathematics

Cite this

The distribution of robust distances. / Hardin, Johanna; Rocke, David M.

In: Journal of Computational and Graphical Statistics, Vol. 14, No. 4, 12.2005, p. 928-946.

Research output: Contribution to journalArticle

@article{adeaa59a4b114b3c85e42d9dd2d9e203,
title = "The distribution of robust distances",
abstract = "Mahalanobis-type distances in which the shape matrix is derived from a consistent, high-breakdown robust multivariate location and scale estimator have an asymptotic chisquared distribution as is the case with those derived from the ordinary covariance matrix. For example, Rousseeuw's minimum covariance determinant (MCD) is a robust estimator with a high breakdown. However, even in quite large samples, the chi-squared approximation to the distances of the sample data from the MCD center with respect to the MCD shape is poor. We provide an improved F approximation that gives accurate outlier rejection points for various sample sizes.",
keywords = "Mahalanobis squared distance, Minimum covariance determinant, Outlier detection, Robust estimation",
author = "Johanna Hardin and Rocke, {David M}",
year = "2005",
month = "12",
doi = "10.1198/106186005X77685",
language = "English (US)",
volume = "14",
pages = "928--946",
journal = "Journal of Computational and Graphical Statistics",
issn = "1061-8600",
publisher = "American Statistical Association",
number = "4",

}

TY - JOUR

T1 - The distribution of robust distances

AU - Hardin, Johanna

AU - Rocke, David M

PY - 2005/12

Y1 - 2005/12

N2 - Mahalanobis-type distances in which the shape matrix is derived from a consistent, high-breakdown robust multivariate location and scale estimator have an asymptotic chisquared distribution as is the case with those derived from the ordinary covariance matrix. For example, Rousseeuw's minimum covariance determinant (MCD) is a robust estimator with a high breakdown. However, even in quite large samples, the chi-squared approximation to the distances of the sample data from the MCD center with respect to the MCD shape is poor. We provide an improved F approximation that gives accurate outlier rejection points for various sample sizes.

AB - Mahalanobis-type distances in which the shape matrix is derived from a consistent, high-breakdown robust multivariate location and scale estimator have an asymptotic chisquared distribution as is the case with those derived from the ordinary covariance matrix. For example, Rousseeuw's minimum covariance determinant (MCD) is a robust estimator with a high breakdown. However, even in quite large samples, the chi-squared approximation to the distances of the sample data from the MCD center with respect to the MCD shape is poor. We provide an improved F approximation that gives accurate outlier rejection points for various sample sizes.

KW - Mahalanobis squared distance

KW - Minimum covariance determinant

KW - Outlier detection

KW - Robust estimation

UR - http://www.scopus.com/inward/record.url?scp=29544436837&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=29544436837&partnerID=8YFLogxK

U2 - 10.1198/106186005X77685

DO - 10.1198/106186005X77685

M3 - Article

AN - SCOPUS:29544436837

VL - 14

SP - 928

EP - 946

JO - Journal of Computational and Graphical Statistics

JF - Journal of Computational and Graphical Statistics

SN - 1061-8600

IS - 4

ER -