Abstract : In many applications using a data base of individual clustering research, we are often led to consider a very important number of qualitative variables. Then the following problem appears : Is it possible to replace a large space of qualitative variables, which is almost impossible to deal with, by a smaller space, containing the same information as the first? The aim of the present article is to solve the problem, classifying qualitative variables with a view to selection and reduction.
Résumé : Il est fréquent d'avoir à effectuer une typologie d'individus à partir d'un très grand nombres de variables qualitatives. On propose dans cet article des méthodes de classification permettant de remplacer un grand ensemble de variables qualitatives, difficile à gérer, par un ensemble plus réduit.