Some clarifications on the DEA clustering approach

This paper clarifies the role of alternative optimal solutions in the clustering of multidimensional observations using data envelopment analysis (DEA). The paper shows that alternative optimal solutions corresponding to several units produce different groups with different sizes and different decision making units (DMUs) at each class. This implies that a specific DMU may be grouped into different clusters when the corresponding DEA model has multiple optimal solutions.

► This paper clarifies some issues of DEA clustering algorithm. ► It emphasizes on the role of alternative optimal solutions in the process of clustering. ► It shows that taking different optimal solutions may give different clustering both in the size and the DMUs within the clusters.