We describe in this paper a method allowing to order submodels in linear regression. A real function is attached to each submodel, allowing to graphically compare and order them. Our procedure defines an objective function depending on two factors (lack of fit and multicolinearity) with the...

Nous etablissons la notion de distance monotone entre un ensemble de variables et une variable opposee a cet ensemble. Cela permet de formaliser le lien entre deux proprietes, la coherence et la stabilite d'unsous-modele de regression et finalement d'etablir un critere de choix de modele base...

Nous nous situons, dans ce papier qui ne fait qu'etablir certains resultats demeurant a exploiter, sur le plan de methodes purement deterministes et presentons une approche qui a trait a la theorie de l'approximation. Nous nous attachons a decouvrir la divergence fondamentale existant entre un...

Aujourd'hui encore. la decision d'un choix d'echange de donnees informatisees (EDI) par le management est freinee par la dificulte de mesurer les atouts de l'entreprise face a ce concept technologique. L'objet de cette etude est de determiner les pre-requis indispensables au succes d'un projet...

In the framework of regression, consider the set of regression submodels. By submodel, we mean on e or more response variables and a subset of the potential regressors. Imagine the submodels as a ponts in some space. How can we "project" these ponts onto a map so as to visualize and compare...

We examine in this paper various consequences of the Poincari separation theorem. First, Poincari's theorem is used to prove results establishing a constrainted principal component analysis, both at the sample and at the population level. Introduction of constraints in principal component...

This article gives results specifically related to the last principal component.

We adapt the 'Almost Ideal Demand System' (AIDS) from consumer expenditure theory to estimate an equation explaining demand for rail passsenger travel. The motivation for using an AIDS is essentially empirical: (1) The model is flexible enough to easily enable the insertion of specific variables...