Finding an unknown number of multivariate outliers

被引：112

作者：

Riani, Marco ^{[2
]}

Atkinson, Anthony C. ^{[1
]}

Cerioli, Andrea ^{[2
]}

机构：

[1] Univ London London Sch Econ & Polit Sci, Dept Stat, London WC2A 2AE, England

[2] Univ Parma, I-43100 Parma, Italy

来源：

JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY | 2009年 / 71卷

关键词：

Forward search; Graphics; Logistic plots; Mahalanobis distance; Minimum covariance determinant; Order statistics; Power comparisons; Simultaneous inference; Truncated distributions; Very robust methods; MULTIPLE OUTLIERS; COVARIANCE; IDENTIFICATION; ASYMPTOTICS; MATRIX; ESTIMATORS; ALGORITHM; POINTS;

D O I：

10.1111/j.1467-9868.2008.00692.x

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

We use the forward search to provide robust Mahalanobis distances to detect the presence of outliers in a sample of multivariate normal data. Theoretical results on order statistics and on estimation in truncated samples provide the distribution of our test statistic. We also introduce several new robust distances with associated distributional results. Comparisons of our procedure with tests using other robust Mahalanobis distances show the good size and high power of our procedure. We also provide a unification of results on correction factors for estimation from truncated samples.

引用

页码：447 / 466

页数：20

共 33 条

[1]

[Anonymous], 1993, Continuous Univariate Distributions, DOI DOI 10.1016/0167-9473(96)90015-8

[2] Exploratory tools for clustering multivariate data [J].