Data envelopment analysis (DEA) is a nonparametric frontier assessment method used to evaluate the relative efficiency of similar decision-making units (DMUs). This method provides benchmarking information regarding the removal of inefficiency. In conventional DEA models, the view of the decision maker (DM) is ignored and the performance of each DMU is solely determined by the observations retrieved. The current paper exploits the structural similarity existing between DEA and multiple objective programming to define a model that incorporates the preferences of DMs in the evaluation process of DMUs. Given the potential unfeasibility of the input and output targets selected by the DM, the model defines an interactive procedure that considers minimum and maximum acceptable objective levels. Given the feasible levels located closer to the targets selected by the DM, a program improving upon the feasible allocations is designed so that the suggested benchmark approximates the requirements fixed by the DM as much as possible. A real-life case study is included to illustrate the efficacy and applicability of the proposed hybrid procedure.

数据包络分析（DEA）是一种非参数前沿评估方法，用于评估类似决策单位（DMU）的相对效率。此方法提供了有关消除效率低下的基准信息。在传统的DEA模型中，决策者（DM）的视图被忽略，每个DMU的性能仅由获取的观察结果确定。本文利用DEA和多目标程序设计之间存在的结构相似性来定义一个模型，该模型将DM的偏好纳入DMU的评估过程中。考虑到DM选择的输入和输出目标的潜在不可行性，该模型定义了一个交互过程，该过程考虑了最小和最大可接受的目标水平。考虑到可行水平更接近DM选择的目标，设计了一个对可行分配进行改进的程序，以使建议的基准尽可能接近DM确定的要求。包括一个实际案例研究，以说明所提出的混合程序的功效和适用性。