INTERPOLATIVE AND PREDICTIVE BIPLOTS APPLIED TO GENERALIZED PLS DISCRIMINANT ANALYSIS

It is possible to show that the duality representation gap for the Generalized PLS Discriminant Analysis (GPLSDA) (Sabatier et al., 2003; Amenta, 2004) is equal to zero and, in this way, the GPLSDA pseudo factorial representations result to be real graphical maps from a duality point of view. However, for inexperienced users of statistics, such outputs are sometimes difficult to interpret. In this context, the GPLSDA representations can be enriched by using the additional relevant information given by the biplot procedures presented by Gower and Hand (1996) to carry out final interpretations. Aim of this paper is to introduce the Gower and Hand‘s biplot procedures within the GPLSDA.