Recent results in ridge regression methods

Autor:	I. B. MacNeill, M. A. Alkhamisi
Rok vydání:	2015
Předmět:	Statistics and Probability Trace (linear algebra) Mean squared error Monte Carlo method Statistics Collinearity Ridge (differential geometry) Condition number Eigenvalues and eigenvectors Regression Mathematics
Zdroj:	METRON. 73:359-376
ISSN:	2281-695X 0026-1424
DOI:	10.1007/s40300-015-0065-4
Popis:	Necessary and sufficient conditions for superiority of the restricted ridge estimator over the restricted least squares estimator are derived when the set of a prior restrictions on parameters are assumed to be incorrect (as well as when the restrictions are assumed to hold). Condition number and trace of mean square error criteria are used to gauge the goodness of some new and some known ridge parameters in rectifying the collinearity problem in three well known real life data sets and a Monte Carlo simulation. Tables 1, 2, 3, 4 and 5 reveal the superiority of the newly formed ridge parameters over some known ridge parameters by means of the foregoing criteria.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::db242a8b557310b8f8a31601bc1c8924 https://doi.org/10.1007/s40300-015-0065-4 Zobrazit plný text záznamu Full text from SpringerLink