ABSTRACT
The axes of the principal component analysis only depend on the correlations between variables; those of the Procrustes analysis depend only on between-table correlations. Multiple Factor Analysis (MFA) and generalised Procrustes analysis are fundamentally different methods with different objectives. The maximum inertia is the same in each group, which eliminates the influence of the weighting of the MFA: both analyses operate on the same data. MFA and generalised Procrustes analysis (GPA) are fundamentally different methods with different objectives. In particular, GPA constructs a representation of the data within a constrained framework. In GPA, the issue is the same as prior to conducting an MFA or any other analysis of this type of multiple table. In MFA, axis by axis, the aim is to maximise projected between-class inertia. MFA is a particular factorial analysis and a particular multicanonical analysis.
