On possibilistic inference (Q2641026)

Given a fuzzy variable, whose possibility distribution function is known up to a parameter, the purpose of this paper is to present ways to estimate the unknown parameter value, on the basis of observations from the variable. Following an analogous path to that in estimation in statistics, the author first adapts the classical maximum-likelihood principle as a possible procedure to estimate the unknown parameter, and later defines a kind of risk function for the new situation, to compare different estimation procedures. It is worth remarking that, in spite of statistical studies on estimation, in those developed in this paper we cannot guarantee an improvement of estimator quality by simply increasing sample size.