| 引用本文: |
姚佳，李俊桥，张青.基于分组的两步DEA基准学习[J].陆军工程大学,2023,(4):32-39
[点击复制] |
|
| |
|
|
| 本文已被：浏览 65次 下载 116次 |
| 基于分组的两步DEA基准学习 |
| 姚佳，李俊桥，张青 |
| (陆军工程大学，江苏 南京 210007) |
| 摘要:数据包络分析（data envelopment analysis, DEA）在为决策单元（decision making unit, DMU）评估效率水平的同时，可为其中的非有效单元提供消除低效的改进措施，即基准信息。但经典DEA模型为非有效单元提供的基准信息不易一步到位，缺乏对分组信息的充分利用。在依赖上下文的DEA框架内进行开发，提出了一种基于分组的两步DEA基准学习模型。模型使用加权L1范式衡量待评估单元与相应目标的接近程度。通过最小化实际点到Pareto有效边界的距离，为每一个决策单元在组内和全局的最佳实践前沿上分别设立单独基准，解决了在实践中目标点难以一步实现的问题，模型的结果可以视为针对最佳实践的长期改进策略。由于充分考虑了分组信息，模型能够反映给定基准过程中涉及的DMU周围环境，并增强了组内DMU在设立目标上的灵活性。该模型被用于评估西班牙公立大学的科研水平，通过对比实验验证了该模型的优势。 |
| 关键词: 数据包络分析 决策单元 基准学习 目标设定 |
| DOI：10.12018/j.issn.2097-0730.20220906001 |
| 投稿时间：2022-09-06 |
| 基金项目:军内科研项目（KYJBJQZL2203） |
|
| Group-Based Two-Step Benchmark Study with DEA |
| YAO Jia，LI Junqiao，ZHANG Qing |
| (Army Engineering University of PLA，Nanjing 210007，China) |
| Abstract: When data envelopment analysis (DEA) is applied to assess the efficiency level of the decision making unit (DMU), it can provide improvement measures, i.e., benchmarking information, to eliminate inefficiencies for non-effective units in it. However, the benchmarking information provided by the classical DEA model for non-effective units is not easy to implement in one step, and lacks the full use of grouping information. This paper develops and proposes a group-based two-step benchmarking model under the context-dependent DEA framework. The model uses the weighted L1 paradigm to measure the proximity between the evaluated DMU and its corresponding target. By minimizing the distance from the actual point to the Pareto boundary, separate benchmarks are established for each DMU on the intra-group and global best practice frontier, which solves the problem that the target point is difficult to achieve in one step in practice. The results of the model can be viewed as a long-term improvement strategy aimed at best practice. In addition, due to the full consideration of grouping information, the model can reflect the surrounding environment of DMU involved in a given benchmarking process and provide flexibility for setting up targets in the group. Finally, the model was used to assess the level of research in Spanish public universities, demonstrating the advantages of this model through comparative experiments. |
| Key words: data envelopment analysis decision making unit (DMU) benchmark study target setting |
|
