基于公共权重改进的情景依赖型DEA模型及应用研究

发布时间：2018-04-27 00:05

本文选题：数据包络分析 + 情景依赖　；参考：《南京邮电大学》2017年硕士论文

【摘要】：随着全球竞争的不断加剧,各行各业相继面临着同质化的问题,质量和价格水平日趋接近。运用传统的的数据包络分析(Data Envelopment Analysis,DEA)模型只能从效率值的角度对决策单元(Decision Making Units,DMUs)进行对比,并且只能区分DMUs是CCR有效还是非有效,不能对CCR有效的DMUs进一步的区分。为了对CCR有效的DMUs进行进一步的分级和排序,Seiford和Zhu构建了情景依赖的DEA模型。传统的情景依赖型DEA(Context-dependent Data Envelopment Analysis,CD-DEA)模型通过计算被评价决策单元与第三方评价单元群体之间的径向距离的方法,来评价决策单元之间的相对吸引力和拓展空间。实际上,这一距离,最终表现为被评价决策单元距离第三方评价单元群体中的某个特定单元的距离,忽略了第三方评价单元群体的综合评价。基于此,Lim提出改进的基于交叉效率的情景依赖型DEA模型,但基于交叉效率改进的情景依赖型DEA模型的交叉吸引力和交叉拓展空间的评价标准是由决策者主观确定的,并且每个DMU对于输出指标的权重都有自己的偏好。基于此,本文在传统情景依赖型DEA模型和基于交叉效率改进的情景依赖型DEA的基础上,提出引入第三方决策单元群体的公共权重,并以此为依据进行被评价单元的吸引力和拓展空间的测度。通过对基于多目标优化的公共权重的DEA模型和基于Dinkelbach算法的公共权重模型的对比分析,最终选择基于Dinkelbach算法的公共权重模型,这一改进保证了第三方评价单元群体决策的一致性。并通过具体的算例对三种模型的计算结果进行了对比分析,表明改进的模型的有效性。
[Abstract]:With the aggravation of global competition, various industries are faced with the problem of homogeneity one after another, and the quality and price level are approaching day by day. The traditional data Envelopment Analysis (DEAA) model can only compare Decision Making units with DMUs from the point of view of efficiency, and can only distinguish whether DMUs is CCR effective or not, and can not be further distinguished from CCR effective DMUs. In order to further classify and sort the effective DMUs of CCR, Seiford and Zhu constructed the DEA model of situational dependency. The traditional situational dependent DEA(Context-dependent Data Envelopment Analysis (CD-DEA) model evaluates the relative attractiveness and expands the space of decision making units by calculating the radial distance between the evaluated decision units and the third party evaluation units. In fact, this distance is the distance between the decision unit under evaluation and a specific unit in the third party evaluation unit group, and the comprehensive evaluation of the third party evaluation unit group is ignored. Based on this, Lim proposes an improved scenario dependent DEA model based on cross efficiency, but the evaluation criteria of cross-attractiveness and cross-expansion space of scenario dependent DEA model based on cross-efficiency improvement are determined by decision makers subjectively. And each DMU has its own preference for the weight of output indicators. Based on the traditional situational dependent DEA model and the scenario dependent DEA based on cross-efficiency improvement, this paper proposes the common weight of the third-party decision unit group. On the basis of this, the attraction of the evaluated unit and the measure of the expansion space are carried out. Through the comparative analysis of the DEA model based on multi-objective optimization and the common weight model based on Dinkelbach algorithm, the common weight model based on Dinkelbach algorithm is selected. This improvement ensures the consistency of the group decision of the third party evaluation unit. The results of the three models are compared and analyzed by a concrete example, which shows the effectiveness of the improved model.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F416.61

【参考文献】