Bibliographic citations
Figueroa, J., (2010). Medidas de diagnóstico para identificar observaciones influyentes en análisis de componentes principales comunes [Tesis, Universidad Nacional Mayor de San Marcos]. https://hdl.handle.net/20.500.12672/2101
Figueroa, J., Medidas de diagnóstico para identificar observaciones influyentes en análisis de componentes principales comunes [Tesis]. PE: Universidad Nacional Mayor de San Marcos; 2010. https://hdl.handle.net/20.500.12672/2101
@misc{renati/474316,
title = "Medidas de diagnóstico para identificar observaciones influyentes en análisis de componentes principales comunes",
author = "Figueroa Agüero, Jeanette",
publisher = "Universidad Nacional Mayor de San Marcos",
year = "2010"
}
-- We present measures to detect and identify influential observations, which have been widely developed in the area of robust and mainly in the context of linear regression models, whose story line, include the work of Belsley (1982), Cook (1986), Atkinson (1986) among others. The Common Principal Component Model by Flury (1984) for several groups of multivariate observations assume that the transformed variables in this model, with major axes equal in all groups but different covariance matrices along common axes groups. In this paper, we present measures for identifying influential observations when the data follow the model of. Is also aligning the diagonal elements of the matrix of local influence of the diagonal elements of the matrix leverage, so allow us to detect sets of observations which coincide simultaneous effects in the identification of these influential observations and illustrated with some applications in botany and agriculture. The method is based on finding a common structure, a rotation (common), which diagonalice covariance matrices of the original data simultaneously in all populations, from the comparison of covariance matrices. The hypothesis for the common basic structure of covariance matrices (positive definite) for populations is: , where: is an orthogonal matrix of eigenvectors, are diagonal matrices of eigenvalues and is the covariance matrix of the th -population.
This item is licensed under a Creative Commons License