摘要
本文讨论属性值用概率分布函数描述的风险型多属性决策问题。首先,分析了利用随机优势描述偏好的局限性;其次,给出概率优势的概念,并得出了随机优势与概率优势在比较随机变量取值优劣方面具有互补性的结论;再次,针对风险型多属性决策问题,结合随机优势与概率优势共同描述方案在单个属性下的局部偏好,并利用赋值级别高于关系的思想集结得到总体偏好关系;最后给出了一个实例,计算结果表明:与文献中仅用随机优势描述偏好的方法相比,本文所给的决策方法的求解结果要好。
This paper investigates the problem of multiple attribute decision making under risk where the consequences of the alternatives are random variables. Firstly, the limitations of stochastic dominance in describing decision maker＇s preference are analyzed. Secondly, the concept of probability dominance is presented. It is pointed out that stochastic dominance and probability dominance can compensate each other in comparing random variables. Thirdly, a decision model based on stochastic dominance and probability dominance for multi-attribute decision making under risk is proposed. In the new model, the global preference is aggregated by the procedure in ELECTRE-Ⅲ method. At last, the method is illustrated by a real example. The results shows that the method presented in this paper can model the decision maker＇s preference more precisely than those in the literature.
出处
《预测》
CSSCI
2007年第3期33-38,共6页
Forecasting
关键词
随机优势
概率优势
风险型多属性决策
赋值级别高于关系
stochastic dominance
probability dominance
multiple attribute decision making under risk
valued outranking relation
