Regresión y análisis factoriales

Some recent developments in factor analysis of multi-sets are introduced in this short course. The more usual factor analyses are based on the singular values de-composition. So, the analyses of two matrices here are introduced from a qualitative criterion, for a non redundancy of partial relations...

Descripción completa

Autor Principal: Lafosse, Róger
Formato: Artículo
Idioma: Español
Publicado: 2015
Acceso en línea: http://revistas.ucr.ac.cr/index.php/matematica/article/view/179
http://hdl.handle.net/10669/12817
Sumario: Some recent developments in factor analysis of multi-sets are introduced in this short course. The more usual factor analyses are based on the singular values de-composition. So, the analyses of two matrices here are introduced from a qualitative criterion, for a non redundancy of partial relations between "common factors". A proposal for extending the simple linear regression, here considered between the two sets of the individuals which define the two sets of variables, lead to specific measures for each matrix. Some juxtapositions of graphics then are justified. The whole previous approach is extended for analyzing the dependency of K matrices with one matrix. The ACOM of Chessel and Hanafi (1996) then is considered as a PCA of sets of variables.Keywords: dependence between sets of variables, non redondancy, simultaneity, biplots, ordinary least squares, correlation, simple linear regression.