Robust high dimensional M-test using regularized geometric median covariance

The original M-test used for testing equality of several independent samples covariance matrices is developed based on likelihood ratio test under assumption of multivariate normality distribution, is sensitive to presence of outliers in the data. The test is fast achieving significant distinction i...

Description complète

Détails bibliographiques
Auteur principal: Kehinde, Alo Olusegun
Format: Thèse
Langue:anglais
Publié: 2018
Sujets:
Accès en ligne:https://etd.uum.edu.my/8528/1/s96163_01.pdf
Description
Résumé:The original M-test used for testing equality of several independent samples covariance matrices is developed based on likelihood ratio test under assumption of multivariate normality distribution, is sensitive to presence of outliers in the data. The test is fast achieving significant distinction in many areas of economics and financial market.