基于变参数对有效决策单元排序性能的研究

Research on Ordering Performance of Effective Decision Making Units Under Different Parameter Variations

随着社会的发展,人们面临不同的决策问题越来越普遍,考虑不同参数变化对DMUs的排序性能的影响更加符合当今时代的意义.以往的DEA研究中,很少有学者改变输入变量的分布来探究超效率模型的“真实”排名的接近程度.基于此,改变输入变量分布来探究DEA与“真实”排名的接近程度.仿真结果表明:1)当n增加或m减少时,AP模型与L-L模型与“真实”排名的概率值均会增加.2)当技术参数u服从半正态分布|N[0,5.06]|,两种模型的排序性能均十分良好,但当技术参数相对无效的时候,两种模型对有效决策单元的排序并不太令人满意.3)当输入变量分布服从U(1,4)时,L-L模型的排序性能要优于AP模型.4)当输入变量分布服从U(1,6)与|N[0,2.5]|时,AP模型的排序性能要优于L-L模型.

With the development of society,it is more and more common for people to face diferent decision problems.Considering the impact of different parameters on the sorting performance of DMUs is more in line with the significance of the modern times.In previous DEA studies,few scholars changed the distribution of input variables to explore how close the super-eficiency model is to the“real"ranking.Based on this,this paper changes the distribution of input variables to explore how close DEA is to the "real"ranking.The sim-ulation results show that:1)Whenn increases or m decreases,the probability value of the"true"ranking of both AP and L-L models will increase.2)When the technical parameters are relatively important,the ranking performance of both models is very good,but when the technical parameters are relatively invalid,the ranking performance of both models is not very satisfactory.3)When the input variable distribution follows U(1,4),the sorting performance of L-L model is better than that of AP model.4)When the distribution of input variables follows U(1,4)and|N[0,2.5]|,the sorting performance of AP model is better than that of L-L model.