Novo estimador de cumeeira de Rao com aplicação em seleção genômica

• Ano de defesa: 2016
• Autor(a) principal: Costa, Luzia Aparecida da
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-Graduação em Estatística e Experimentação Agropecuária; UFLA; brasil; Departamento de Ciências Exatas
• Palavras-chave em Português: Análise Multivariada; Estimadores viesados; Multicolinearidade; Ridge Regression; Multivariate Analysis; Biased estimators; Multicollinearity; Probabilidade e Estatística Aplicadas
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/11100

A new ridge estimator is obtained from a geometric approach of the general theory of the ridge estimators. Obtained from the intersection of two orthogonal ellipsoids and having a simple algebraic expression this estimator is quite natural. Examples are used to suggest that curves ridge from this estimator stabilize faster than the usual ridge estimators. This fact may facilitate the estimation of great value for the parameter, a central area of research in theory. Expressions were obtained for the total variance and the mean quadratic error. The performance of the proposed estimator was studied by applying a set of data already analyzed in the literature by using the usual ridge estimator. Data analysis of genomic selection on the pH response of pork at 45 minutes and 24 hours after slaughter. In all applications the results obtained by the proposed estimator are compared with the results obtained by the usual ridge and Mayer and Wilke (1973) estimator.