Résumé (EN)

Each year, Régie Autonome des Transport Parisiens RATP renovates a certain number of metro stations. The method used to determine which ones should be chosen for renovation, was reconsidered two years ago. A programming system, using updated data, now ranks by priority each year all the stations renovated. The ranking procedure uses Electre III. Each station is evaluated using seven criteria arising from the objectives and constraints of either the passengers or the RATP. For each criterion, indifference and preference thresholds are determined after establishing the relative importance of each criterion. A joint research contract was drawn up by Laboratoire d'Analyse et Modelisstion de Systémes pour l'Aide à la Décision (LANSADE, University of Paris) and, RATP for the implementation of the method. This study led to two critical developments in multicriteria methods: 1. — A method for obtaining the necessary information from RATP staff to quantify the relative importance of each criterion was rested; it led to the determination of the variation ranges of criteria ‘weights’. 2. — A method to appraise the impact of imperfect information — which can take several forms — on the ranking was developed; it was applied to 32 sets of values.