Abstract

Data envelopment analysis (DEA) is a methodology for evaluating efficiencies of decision-making units (DMUs) with each unit having its own set of inputs and outputs. However, there are situations where there can be an interdependence among the units. In a previous paper the authors examine efficiency measurement in a situation where university departments are grouped by faculty and share a single resource at the faculty level. Furthermore, the shared resource is assumed to be one which cannot be split up and allocated to the group members. The current paper generalizes that earlier work by considering decision-making units grouped according to multiple attributes and with multiple shared inputs. In addition, the problem of overlapping groups is investigated. A DEA-like methodology is developed for deriving efficiency scores in this multiple attribute situation. Further, we present a methodology for evaluating efficiency at the level of the groups, e.g. the level of the faculty, as well as at the level of the members within the groups. To further demonstrate the need for such methodologies, we present a number of real-world problem settings where shared factors and groupings of DMUs need to be dealt with.

摘要

数据包络分析 (DEA) 是一种评估决策单元 (DMU) 效率的方法，每个单元都有自己的一组输入和输出。但是，在某些情况下，单元之间可能存在相互依赖关系。在之前的一篇论文中，作者研究了大学部门按教师分组并在教师级别共享单一资源的情况下的效率测量。此外，假设共享资源是一种不能被分割和分配给组成员的资源。当前的论文通过考虑根据多个属性和多个共享输入分组的决策单元来概括早期的工作。此外，研究了重叠组的问题。开发了一种类似 DEA 的方法，用于在这种多属性情况下得出效率分数。此外，我们提出了一种在组，例如教师级别，以及组内成员级别。为了进一步证明这种方法的必要性，我们提出了一些现实世界的问题设置，其中需要处理共享因素和 DMU 分组。