Ranking fuzzy interval numbers in the setting of random sets (Q1803193): Difference between revisions

It may be related to any fuzzy interval number \(G\) a class of random sets, say \(RS(G)\), which are equivalent to \(G\) ``under one-point coverages''. Any random set \(S\to RS(G)\) fulfils the following condition: \(\text{Prob}(x\in S)=\mu_ G(x)\), \(x\in R\), where \(\mu_ G\) is a membership function of \(G\). Among the random sets in \(RS(G)\) some interval-valued sets are of special interest. They are: random level set of \(G\), asymmetric random level set of \(G\) and random interval set of maximum entropy. These random sets may be used for determining fuzzy order relations in the set of fuzzy interval numbers. Let \(\geq\) be any order relation determined in set \(I\) of all closed intervals. Then it may be extended into a fuzzy relation \(\geq_ f\) given in the set \(FI\) of all fuzzy internal numbers by defining membership function \(\mu_{\geq_ f}:FI\times FI\to[0,1]\) in the following way: \(\mu_{\geq_ f}(A,b)=\text{Prob}(S_ A\geq S_ B)\), \(A,B\in FI\), where \(S_ A\) and \(S_ B\) are fixed random sets: \(S_ A\in RI(A)\) and \(S_ B\in RI(B)\). Depending on relation \(\geq\), random sets \(S_ A\) and \(S_ B\) as well as on the type of dependence between \(S_ A\) and \(S_ B\) assumed in the above formula one may obtain different fuzzy order relations useful for ranking fuzzy interval numbers. In the paper eight of them are considered. Some properties of these relations are also investigated and exact formulae for calculating \(\mu_{\geq_ f}(A,B)\) for trapezoidal fuzzy numbers \(A\) and \(B\) are derived.