A special measure of uncertainty (Q756623)

The paper introduces a measure of fuzziness which is used to evaluate a level of hesitation (``hardness'') of decision-making in presence of several highly competitive choices. Let A be a fuzzy set of decision, A: \(X\to [0,1]\). The main objective of the proposed measure is to provide a clear distinction between the following two decision situations: (i) Two or more grades of membership of A are equal to 1, say \(A(x_ 1)=A(x_ 2)=1\), \(x_ 1\neq x_ 2\) (the decisions \(x_ 1\) and \(x_ 2\) are equivalent); (ii) for \(x_ 1\) and \(x_ 2\) one has \(A(x_ 1)=1\), \(A(x_ 2)=1-\epsilon\), where \(\epsilon\) stands for a small positive number. This decision situation gives rise to a significant level of hesitation. Departing from the measure of uncertainty defined as \[ U(A)=\int^{1}_{0}\log_ 2(card A^{\alpha})d\alpha \] the new measure uses the modified cardinality so that the cases (i) and (ii) are distinguishable. A simple example coming from quadratic programming provides an appropriate numerical illustration.