A biased-robust regression technique for the combined outlier-multicollinearity problem

James R. Simpson, Douglas Montgomery

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

The combined outlier-multicollinearity problem occurs frequently in regression data. Methods that successfully address this problem effectively combine biased and robust estimation techniques. A biased-robust estimator is proposed that uses a multi-stage generalized M-estimator with fully iterated ridge regression to successfully control both influence and collinearity in regression datasets. Two previously published approaches are compared with the proposal via simulation experiments. The best performing published technique is also compared with our proposal using a dataset containing a cloud of outliers and severe multicollinearity. The proposed biased-robust method outperforms the published technique both in simulation and the example.

Original languageEnglish (US)
Pages (from-to)1-22
Number of pages22
JournalJournal of Statistical Computation and Simulation
Volume56
Issue number1
StatePublished - 1996

Fingerprint

Multicollinearity
Robust Regression
Outlier
Biased
Regression
Biased Estimation
Collinearity
Ridge Regression
M-estimator
Robust Estimators
Experiments
Robust Estimation
Robust Methods
Simulation Experiment
Outliers
Robust regression
Simulation

Keywords

  • Bounded influence
  • Breakdown point
  • Efficiency
  • Montecarlo simulation
  • Ridge regression

ASJC Scopus subject areas

  • Applied Mathematics
  • Modeling and Simulation
  • Statistics and Probability

Cite this

A biased-robust regression technique for the combined outlier-multicollinearity problem. / Simpson, James R.; Montgomery, Douglas.

In: Journal of Statistical Computation and Simulation, Vol. 56, No. 1, 1996, p. 1-22.

Research output: Contribution to journalArticle

@article{c4ac313d913e415e8ffa963acb7675ae,
title = "A biased-robust regression technique for the combined outlier-multicollinearity problem",
abstract = "The combined outlier-multicollinearity problem occurs frequently in regression data. Methods that successfully address this problem effectively combine biased and robust estimation techniques. A biased-robust estimator is proposed that uses a multi-stage generalized M-estimator with fully iterated ridge regression to successfully control both influence and collinearity in regression datasets. Two previously published approaches are compared with the proposal via simulation experiments. The best performing published technique is also compared with our proposal using a dataset containing a cloud of outliers and severe multicollinearity. The proposed biased-robust method outperforms the published technique both in simulation and the example.",
keywords = "Bounded influence, Breakdown point, Efficiency, Montecarlo simulation, Ridge regression",
author = "Simpson, {James R.} and Douglas Montgomery",
year = "1996",
language = "English (US)",
volume = "56",
pages = "1--22",
journal = "Journal of Statistical Computation and Simulation",
issn = "0094-9655",
publisher = "Taylor and Francis Ltd.",
number = "1",

}

TY - JOUR

T1 - A biased-robust regression technique for the combined outlier-multicollinearity problem

AU - Simpson, James R.

AU - Montgomery, Douglas

PY - 1996

Y1 - 1996

N2 - The combined outlier-multicollinearity problem occurs frequently in regression data. Methods that successfully address this problem effectively combine biased and robust estimation techniques. A biased-robust estimator is proposed that uses a multi-stage generalized M-estimator with fully iterated ridge regression to successfully control both influence and collinearity in regression datasets. Two previously published approaches are compared with the proposal via simulation experiments. The best performing published technique is also compared with our proposal using a dataset containing a cloud of outliers and severe multicollinearity. The proposed biased-robust method outperforms the published technique both in simulation and the example.

AB - The combined outlier-multicollinearity problem occurs frequently in regression data. Methods that successfully address this problem effectively combine biased and robust estimation techniques. A biased-robust estimator is proposed that uses a multi-stage generalized M-estimator with fully iterated ridge regression to successfully control both influence and collinearity in regression datasets. Two previously published approaches are compared with the proposal via simulation experiments. The best performing published technique is also compared with our proposal using a dataset containing a cloud of outliers and severe multicollinearity. The proposed biased-robust method outperforms the published technique both in simulation and the example.

KW - Bounded influence

KW - Breakdown point

KW - Efficiency

KW - Montecarlo simulation

KW - Ridge regression

UR - http://www.scopus.com/inward/record.url?scp=0030362143&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030362143&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0030362143

VL - 56

SP - 1

EP - 22

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 0094-9655

IS - 1

ER -