A computationally efficient approach to ranking fuzzy numbers

Fuzzy numbers have become very important in decision making system. They allow an agent to provide opinions or recommendations while representing the uncertainty in that opinion. Unfortunately, fuzzy numbers cannot be directly compared as can crisp numbers. Numerous methods have been proposed to facilitate the comparison and ranking of fuzzy numbers. Many of these methods are computationally complex, precluding their use in systems requiring a real-time response. Others methods provide ranking that does not correspond to human intuition. A common problem is the inability to discriminate between certain fuzzy numbers, such as those with identical modes and similar spreads. This paper presents new fuzzy preference function that provides ranking corresponding to human intuition. The method provides the ability for a decision maker to specify preferences for larger or smaller spreads and for the left or right spread of fuzzy numbers. This provides flexibility for various applications. It is also computational efficient.