Multi-criteria decision analysis with fuzzy pairwise comparisons

The authors present a modification of the fuzzy multi-criteria method proposed by Van Laarhoven and Pedrycz. First the weights of the decision criteria are calculated by the minimization of a logarithmic regression function. Next the weights of the decision alternatives are calculated for each criterion separately. Lastly, the fuzzy final scores of the alternatives are determined by an appropriate aggregation of the calculated weights. The authors will show that the present modified method, in contrast with the original one, yields weights which are optimal with respect to the logarithmic regression function. They also present a version of the method with geometric ratio scales, rather than Saaty’s equi-distant scale, to quantify the gradations of human comparative judgment. This approach also enables them to develop an efficient procedure for investigating the scale sensitivity of the calculated weights and the final scores of the alternatives.