Processus interactif d'optimisation avec prise en charge des préférences de l'utilisateur

Authors: Gauthier, Alexis
Advisor: Gaudreault, Jonathan
Abstract: A decision maker using an optimization system may get a solution that he considers inappropriate. It is possible for him, if he uses an interactive reoptimization system, to add a constraint or a preference and to require a new solution. However, preferences for the various values composing the solution are usually lost over the iterations. This thesis proposes an approach for taking into account the user’s preferences. It uses mathematical goal programming techniques. A methodology for implementing the approach is also proposed. Finally, a comparison is made between the proposed approach and another one using heuristics to solve the problem of interactive planning of contributions and withdrawals from a Registered Education Savings Plans. In both cases, the prototypes make it possible to adjust, in real time, and from the tip of the mouse, the solutions on interactive graphics. The prototype, moved by specific heuristics, does not allow the user to reach all admissible solutions. This is often caused by circular adjustments problems where the user may reach a previous state after some iterations. The prototype using mathematical goal programming allows the user to navigate coherently through the solution space. In most contexts, this method should make it easier for the decision maker to access his preferred solution.
Document Type: Mémoire de maîtrise
Issue Date: 2019
Open Access Date: 22 October 2019
Grantor: Université Laval
Collection:Thèses et mémoires

Files in this item:
Description SizeFormat 
35336.pdf7.52 MBAdobe PDFThumbnail
All documents in CorpusUL are protected by Copyright Act of Canada.