Forecasting techniques using fuzzy set theory

Summary form only given. In this paper, we studied Delphi method, exponential smoothing, moving-average, and least-squares method by using fuzzy set theory. The procedure followed in Delphi method is to solicit and collate opinions about a certain subjects from experts and feed back digested appraisals to narrow the differences among opinions until a near agreement is obtained. Kaufmann describes a Delphi method using triangular fuzzy numbers. In his didactic example, Kaufmann considers a technological progress well specified and makes an inquiry with twelve experts. Every expert gives the earliest date, the date of maximum of presumption, and the latest date for the technological progress considered. Then, the mean and the deviations from the mean are computed. The experts are informed and return their respective values with the possibility of some changes. The process continue until the analysts find a stability. In exponential smoothing method, a weighting factor that gives more or less prominence to recent happenings is selected. The moving-average method predicts the demand for the next period from the average demand experienced during several recent periods. The least-square method can be used to determine the line of best fit to plotted data points. In the case of fuzziness, we use a sequence of interval of confidence instead of a sequence of ordinary numbers of actual demand. This is because we assume that we don´t have exact and reliable data of old demand to use in the least-squares method