DMFA: Dual Multiple Factor Analysis

Abstract

In this article, we propose a new method called Dual Multiple Factor Analysis (DMFA), which is an extension of DMFA in the case where individuals are structured according to a partition. The heart of the method rests on a factor analysis known as internal, in reference to the internal correspondence analysis, for which data are systematically centered by group. This analysis is an internal PCA when all the variables are quantitative. DMFA provides the classic results of a PCA as well as additional outputs induced by the consideration of a partition on individuals, such as the superimposed representation of the L scatter plots of variables associated with the L groups of individuals and the representation of the scatter plot of the correlations matrices associated each one with a group of individuals.