Abstract Practically, systems contain a set of subunits each of which has a set of inputs and produces a set of outputs. Conventional data envelopment analysis (DEA) ignores the operations of individual subunits in a system and considers the system as a black-box. Recent DEA studies developed to investigate the internal structures of a system and examine the interrelationship of the components within a system either with serial or parallel or mixed structure. Nevertheless, in many practical settings the assumption of identical inputs and outputs may not hold among the components. As an example, the case of a set of plants, each has various product lines manufactured by different inputs and outputs captures the idea. This paper proposes a network DEA-based methodology to address the problem of nonhomogeneity in settings where subunits operate in the mixed network structure. The layers and the overall system efficiencies can be evaluated through our proposed methodology. To show the applicability of our models, we then use the models and examine the efficiencies of a set of hypothetical data set.

中文翻译：

---

非均匀混合网络的DEA

摘要实际上，系统包含一组子单元，每个子单元具有一组输入并产生一组输出。常规数据包络分析（DEA）忽略了系统中各个子单元的操作，并将系统视为黑匣子。最近的DEA研究发展为调查系统的内部结构，并检查系统中具有串行或并行或混合结构的组件之间的相互关系。但是，在许多实际设置中，组件之间可能不存在相同的输入和输出的假设。例如，在一组工厂的情况下，每个工厂都有由不同的输入和输出制造的各种产品线，从而抓住了这个想法。本文提出了一种基于网络DEA的方法，以解决子单元在混合网络结构中运行的环境中的非均匀性问题。可以通过我们提出的方法来评估各层和整个系统的效率。为了展示我们模型的适用性，我们然后使用模型并检查一组假设数据集的效率。