基于定性指标价值位置概率估计的效率模型

Efficiency Model Based on Value-Position Probability-Estimation of DMUs on Qualitative Criteria

为克服CKS DEA(CookKressSeiford DataEnvelopmentAnalysis)模型所采用的价值排序结构没有能力合理界定和控制决策单元之间的价值差异的缺陷,通过引入对决策单元定性指标偏好及其所处价值排序位置的概率估计,提出了一种测度决策单元相对效率的模型方法。在定性指标价值排序上,该方法采取等距价值结构,改进了CKS DEA模型所依赖的非等距价值结构。等距价值结构的引入,使决策者或有关专家对决策单元定性指标价值偏好位置进行概率估计成为了可能;更有助于分析者通过价值位置的概率估计,构造出约束DEA权重取值的合理置信域,从而提高决策单元相对效率评价的分辨性。对10个决策单元进行效率评价的实例应用表明,利用该模型评价出的效率单元只有4个,而若用CKS DEA模型进行评价则会有9个单元被评为效率单元,从而证实了提出的新模型方法具有比CKS DEA模型更好的相对效率评价分辨能力。

The CKS-DEA(Cook Kress Seiford-Data Envelopment Analysis) model is not able to make well control on the value (preference) difference of qualitative inputs and outputs among DMU (Decision Making Units). To overcome this drawback, this paper presents a new efficiency evaluation model by incorporating decision-makers' preference estimates on qualitative inputs and outputs of DMUs, and the corresponding VPPE (Value-Position Probability-Estimation) on preference estimates. The model adopts an ESWS (Equally-Spaced Weight Structure) instead of the UESWS. The ESWS makes it possible to incorporate into the new model the VPPEs on preference estimates, and through VPPEs more reasonable assurance regions can be constructed, which is more helpful for the given model to differentiate DMUs. Applied to the relative efficiency evaluation problem of 10 real DMUs, the presented model only differentiates 4 DMUs as efficient DMUs, while by the referred model, i.e., the CKS-DEA model, 9 DMUs are evaluated as efficient, which proves that the presented model is more effective in differentiating DMUs than the original CKS-DEA model.