Fuzzy hierarchical analysis revisited (Q5925936): Difference between revisions

The authors suggest a new method for carrying out fuzzy hierarchical analysis. The method is based on the direct fuzzification of Saaty's method for crisp hierarchical analysis. The main idea consists in finding the crisp solution, i.e. crisp vector of weights as a unique, positive, normalized eigenvector corresponding to the maximal eigenvalue of the positive reciprocal matrix \(A\). For finding this vector, Saaty's method is used. There exist formulae expressing the weights as crisp functions of elements of \(A\) for matrices of order 3 and 4. In the article, these formulae are fuzzified by substituting the elements of \(A\) by trapezoidal-type fuzzy numbers and making use of the extension principle. As a result, fuzzy weights for hierarchical analysis are obtained. This new method avoids a complicated direct computation of fuzzy eigenvalues and eigenvectors of a fuzzified positive reciprocal matrix. Some test numerical examples are solved, an application is shown and the consistency of fuzzy hierarchical analysis is discussed.