The fuzzy Delphi method via fuzzy statistics and membership function fitting and an application to the human resources

A new fuzzy Delphi method is proposed. The method employs the fuzzy statistics and technique of the conjugate gradient search to fit membership functions. Membership functions besides triangles may be derived for the fuzzy forecasts. A procedure termed as stabilization rather than convergence of the iterative interval-valued surveys is utilized, in which the alpha-level set of the fuzzy forecasts is used. The testing of stability of the process is conducted on the fuzzy forecasts between two consecutive iterations instead of within each iteration. An application of this method to the problem of managerial talent assessment for a company located in Taiwan and the obtained results are provided. (C) 2000 Elsevier Science B.V. All rights reserved.