A HYBRID APPROACH BASED ON A NEW OUTRANKING HYPOTHESIS COMBINING THE AHP AND ELECTRE II METHODS FOR MULTI-CRITERIA DECISION-MAKING

---

Résumé

Multi-criteria decision-making is crucial in many fields, as decision makers must evaluate a variety of criteria that are sometimes
contradictory. Among the methods widely used to structure and solve complex problems are the analytic hierarchy process (AHP) and the
ELECTRE II method (elimination and choices translating reality). However, each of these methods has its limitations. AHP is effective
for weighting criteria, but it is compensatory in nature, allowing one alternative to compensate for the weakness of one criterion with the
strength of another. Conversely, ELECTRE II, a non-compensatory method, is excellent for managing preferences using concordance and
discordance indices. However, it can lack flexibility in adjusting weightings and concordance and discordance thresholds. To overcome
these limitations, this work proposes a hybrid method, the hybrid AHP-ELECTRE II approach (APHY-AHP-ELECTRE II), which
combines the advantages of AHP and ELECTRE II by relying on the principles of concordance-dominance and discordance-dominance as
ranking assumptions. This hybridization improves the accuracy and robustness of multi-criteria decisions by integrating the weighting of
criteria using AHP. The resulting model is balanced and flexible, and particularly suited to problems where many criteria are in conflict.
This document presents a clear methodology for the different stages of this hybrid method, explores the properties of the outranking
assumption and demonstrates its effectiveness using a numerical application. The integration of the advantages of analytic hierarchy
process (AHP) and ELECTRE II makes the APHY-AHP-ELECTRE II hybrid method flexible and robust decision-making tool, capable of
effectively responding to the challenges of complex decision-making environments.

Mots-clés

APHY-AHP-ELECTRE II, concordance-dominance, discordance dominance, weighting, thresholds, robustness.