Abstract
A problem on estimation of a membership function is formulated as to estimate the expected value of a fuzzy random variable. This formulation makes it possible to discuss the estimation problem objectively. First it is shown that fuzziness can be estimated when information of individual membership functions is available. Secondly a statistical method is proposed for estimation and applied to a practical example. Moreover it is pointed out that a method based on frequency obtained from repeated measurement, one of the former methods, cannot be demonstrated its reasonability.
