Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A local influence approach to identifying multiple multivariate outliers.

W Y Poon1, S F Lew, Y S Poon

  • 1Department of Statistics, Chinese University of Hong Kong, Shatin, Hong Kong. wypoon@hp735.sta.cuhk.edu.hk

The British Journal of Mathematical and Statistical Psychology
|December 8, 2000
PubMed
Summary

This study introduces a novel geometrical approach for detecting multivariate outliers, unifying existing measures. The method offers a flexible and robust alternative for complex data scenarios, including small samples and non-normal distributions.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Portomesenteric vein thrombosis following laparoscopic sleeve gastrectomy in a Chinese patient.

Hong Kong medical journal = Xianggang yi xue za zhi·2017
Same author

Extrapulmonary involvement associated with Mycoplasma pneumoniae infection.

Hong Kong medical journal = Xianggang yi xue za zhi·2015
Same author

The first novel influenza A (H1N1) fatality despite antiviral treatment and extracorporeal membrane oxygenation in Hong Kong.

Hong Kong medical journal = Xianggang yi xue za zhi·2009
Same author

Conditional local influence in case-weights linear regression.

The British journal of mathematical and statistical psychology·2001
Same author

Acute renal failure in a healthy young adult after dextran 40 infusion for external-ear reattachment surgery.

British journal of plastic surgery·2000
Same author

Bayesian analysis of square ordinal-ordinal tables.

The British journal of mathematical and statistical psychology·1999

Area of Science:

  • Statistics
  • Data Analysis

Background:

  • Classical outlier detection methods often rely on distributional assumptions and large sample sizes.
  • Existing measures for multivariate outliers can be complex and lack unification.

Purpose of the Study:

  • To develop new measures for detecting multivariate outliers.
  • To unify existing outlier identification measures using geometrical concepts.
  • To provide a flexible and robust approach for outlier detection in complex situations.

Main Methods:

  • Utilizing Cook's local influence approach and its modification by Poon and Poon.
  • Developing geometrical measures for multivariate outlier detection.
  • Transforming classical measures to the unit interval for unification.

Related Experiment Videos

Main Results:

  • The proposed measures exhibit a form similar to existing literature but offer advantages.
  • The new approach unifies outlier identification measures.
  • The method is independent of distributional assumptions and large-sample properties.

Conclusions:

  • The developed approach provides a valid reason for using various measures in complicated situations.
  • This method is suitable for non-normal cases and small-sample problems.
  • Offers flexibility in identifying outliers with respect to different metrics.