区间估计:AHP指标筛选的一种方法

Range Estimation Priority: A Method to Sift Out Weak-weighted Criteria in AHP

分析了弱权重指标对层次分析法判断矩阵不一致性的负面影响,提出了识别和剔除弱权重指标的基本原理,以及利用层次分析的区间估计剔除弱权重指标的步骤、方法.它尤其适合于决策者希望通过借鉴一个较大的一般指标集建立具体评价问题的指标集的情形.弱权重指标的剔除有助于提高决策者权重比较的精确度,决策者也不再需要收集各方案在那些弱权重上的表现信息,从而提高了层次分析法的应用性.

In this paper, it is shown priority vectors are full of deviations and uncertainties. This occurs due to the decision-makers＇ uncertainties and errors in comparisons as well as inconsistencies in judgment matrices in Analytical Hierarchy ProFess （AHP）. The contribution to evaluation goals of the criteria with very weak weights may be submerged in such deviations and uncertainties, which makes those criteria meaningless and should be sifted out. The concept and method to sift such criteria out are proposed in this paper. This method allows decision-makers to identify and sift out weak-weighted criteria; this is especially helpful for decision-makers who want to build a set of criteria for a specific problem from a large set of criteria for general purposes. With downsizing in criteria numbers, decision- makers are expected to make fewer errors in judgment, and it＇s easier for judgment matrices to reach consistency. Moreover, the performance information of each alternative on those weak-weighted criteria is no more necessary, which further improves the usability of AHP. It＇s proved that this method is feasible, convenient and robust by the numerical case provided in this paper.