关于层次分析法的数值试验

Numerical Experiment on the Analytic Hierarchy Process

通过对层次分析法进行数值模拟,证实了次序性程度不仅对评价结果存在着显著的影响,而且也与一致性指标存在着密切的关系,同时也证实了标度精度不足是导致多种标度并存的原因,并建议采用适当的标度精度,这一观点与Saaty是一致的。实验结果还清楚地表明,单纯使用一致性指标判定一个矩阵好坏是不够的,应该辅助以其它指标。

This paper proves through a numerical experiment on the analytic hierarchy process that the degree of order not only has an apparent influence on the evaluation result, but also bears a close relationship to the coincidence indicator. It is also proves that the lack of scale and precision is the cause of the coexistence of multiple scales with a recommended proper scale and precision. This opinion is consistent with that proposed by Saaty. The experimental results of this study clearly show that using only the coincidence indicator to decide whether a matrix is good is not adequate and should be complimented by other indexes.