Apport de la Méthodologie de la Recherche Expérimentale àun modèle de la température de transition vitreuse de polymères acryliques et méthacryliques

D-optimal design has been used for constructing a experimental database required to set up a high quality model. We show that using a priori statistics criteria (i.e., before making the experiences) it is possible to select the minimum number of molecules required to set up the database. Using the same tools, and the previous results, it is possible to evaluate the quality of the predicted value for a set of new candidates and decide whether or not they have to be taken into account in the database.