Characteristic evaluation for groups in data envelopment analysis and its application

To clarify the usage of Multi-Frontier DEA (Data Envelopment Analysis), this paper presents how to interpret the result computationally. As first challenge, the proposed method is prepared for two groups, and indicates how to characterize each group. The analyzing process consists of two steps: (1) Classify DMU (Decision Making Unit) into one of five categories and (2) Identify representative DMU in a group and find the characteristics in virtual inputs and outputs. Applying this method to two-stage bank model with six attributes, this paper also characterizes the difference of bank efficiency between mega bank group and small bank group.