Abstract

Multi-Criteria Decision Making (MCDM) is a branch of operation research that aims to empower decision makers (DMs) in complex decision problems, where merely depending on DMs judgment is insufficient. Conventional MCDM approaches assume that precise information is available to analyze decision problems. However, decision problems in many applications involve uncertain, imprecise, and subjective data.

This manuscripts-based thesis aims to address a number of challenges within the context of MCDM under uncertain evaluations, where the available data is relatively small and information is poor. The first manuscript is intended to handle decision problems, where interdependencies exist among evaluation criteria, while subjective and objective uncertainty are involved. To this end, a new hybrid MCDM methodology is introduced, in which grey systems theory is integrated with a distinctive combination of MCDM approaches. The emergent ability of the new methodology should improve the evaluation space in such a complex decision problem.

The overall evaluation of a MCDM problem is based on alternatives evaluations over the different criteria and the associated weights of each criterion. However, information on criteria weights might be unknown. In the second manuscripts, MCDM problems with completely unknown weight information is investigated, where evaluations are uncertain. At first, to estimate the unknown criteria weights a new optimization model is proposed, which combines the maximizing deviation method and the principles of grey systems theory. To evaluate potential alternatives under uncertain evaluations, the Preference Ranking Organization METHod for Enrichment Evaluations approach is extended using degrees of possibility.

In many decision areas, information is collected at different periods. Conventional MCDM approaches are not suitable to handle such a dynamic decision problem. Accordingly, the third manuscript aims to address dynamic MCDM (DMCDM) problems with uncertain evaluations over different periods, while information on criteria weights and the influence of different time periods are unknown. A new DMCDM is developed in which three phases are involved: (1) establish priorities among evaluation criteria over different periods; (2) estimate the weight of vectors of different time periods, where the variabilities in the influence of evaluation criteria over the different periods are considered; (3) assess potential alternatives.