Fuzzy multi-criteria decision-making

Multi-criteria decision making (MCDM) models, like the Analytic Hierarchy Process and the Non-Traditional Capital Investment Criteria, have been developed to incorporate the difficult-to-quantify criteria into the decision making process. One shortcoming of these methods is the failure to realistically represent the imprecision of the decision makers' judgments. In this aspect, the use of fuzzy logic and linguistic variables have attracted some attention.; Due to the advantages of the Non-Traditional Capital Investment Criteria (NCIC) over the Analytic Hierarchy Process (AHP), we extended NCIC to include fuzzy logic. We studied the mathematical relationship between AHP and NCIC as a foundation for the development of the fuzzy NCIC methodology. We generated equations to map the outcome of one method into a set of data comparable with the outcome of the other method and demonstrated their use through an application to a real world decision making problem.; A problem well-documented in the use of fuzzy pairwise comparisons in decision models is that of the irrational fuzzy weight vector. We investigated the cause of obtaining irrational weight vectors when pairwise comparisons matrices contained Triangular Fuzzy Numbers as decision makers' judgments. We determined the underlying cause of this phenomenon and developed a testing procedure that will prevent these irrationalities from occurring in the weight vectors.; We extended the strong transitivity axiom to fuzzy pairwise comparisons matrices and developed a method to check for the internal consistency of decision makers when judgments are presented as fuzzy numbers. We studied the relationships between the irrationality of the weight vectors, internal consistency of the decision makers and the satisfaction of the weak monotonicity axiom of measurement theory.; To have a complete fuzzy MCDM method, the use of three different instruments for eliciting judgments from the decision makers in making pairwise comparisons are evaluated. The problem of representing the imprecision as realistically and correctly as possible is addressed by comparing these three instruments through an experimental design using a large panel of subjects.