Support for awareness of data selection in Data Envelopment Analysis

To measure efficiency of DMUs (Decision Making Units), DEA (Data Envelopment Analysis) clarifies relations between input and output elements respectively. Although selection of these data is entrusted to analyst, criteria of selection are not universal. Therefore, reasonability of data selection is sometimes controversial. Then the purpose of this study is to obtain input and output elements based on analyst´s intention for making specific DMUs efficient. It aims for playing an aspect in the discriminant analysis. The traditional method does not focus on the number of minimum necessary elements and procedure is complex. Then this paper proposes method for obtaining optimum elements effectively by considering the less number of elements. Numerical experiments have shown effectiveness of the proposed method.