A heuristic method for choosing 'virtual best' DMUs to enhance the discrimination power of the augmented DEA model

Document Type : Article


School of Industrial Engineering, Iran University of Science & Technology, Tehran, Iran


Despite its intrinsic advantageous features as a tool for increasing discrimination power of the basic DEA (data envelopment analysis) model, augmented DEA has two main drawbacks including the presence of unrealistic efficiency scores and the presence of great distance between its efficiency scores and scores obtained by primary model. In this regard, this paper extends a heuristic method for dealing with both issues and improving the power of augmented DEA model in performance evaluation. Since different virtual DMUs lead to different results for ranking, the hierarchical clustering algorithm is applied in this study to select the best virtual DMUs in order to reduce the possibility of having inappropriate efficiency scores. Finally, to demonstrate the superiority of the proposed approach over previous approaches in literature, two numerical examples are provided.