FUZZY ORDERING
Decisions are sometimes made on the basis of rank, or ordinal ranking: which issue is best, which is second best, and so forth. For issues or actions that are deterministic, such as y1 =5, y2 =2, y1 ≥ y2, there is usually no ambiguity in the ranking; we might call this crisp ordering. In situations where the issues or actions are associated with uncertainty, either random or fuzzy, rank ordering may be ambiguous. This ambiguity, or uncertainty, can be demonstrated for both random and fuzzy variables. First, let us assume that the uncertainty in rank is random; we can use probability density functions (pdf) to illustrate the random case. Suppose we have one random variable, x1, whose uncertainty is characterized by a Gaussian pdf with a mean of μ1 and a standard deviation of σ1, and another random variable, x2, also Gaussian with a mean of μ2 and standard deviation of σ2. Suppose further that σ1 >σ2 and μ1 >μ2. If we plot the pdfs for these two random variables in Figure 9.1, we see that the question of which variable is greater is not clear. As an example of this uncertain ranking, suppose x1 is the height of Italians and x2 is the height of Swedes. Because this uncertainty is of the random kind, we cannot answer the question “Are Swedes taller than Italians?” unless we are dealing with two specific individuals, one each from Sweden and Italy, or we are simply assessing μ1, average-height Swedes, and μ2, average-height Italians. But we can ask the question, “How frequently are Swedes taller than Italians?” We can assess this frequency as the