Group DEA for building consensus by DMU efficiency evaluation

This paper proposes a new method for building group consensus, focusing on evaluating systems with the observed values of the multiple inputs and outputs. Here, the systems mean Decision Making Units (DMUs) in Data Envelopment Analysis (DEA). This method is composed of two parts. In the first part, it evaluates the relative efficiencies of DMUs by group members with DEA. Next, by using weights of the DEA solutions, the authors define and calculate the mutual evaluation information of group members. In the second part, they give the method for making a group evaluation which is expressed as the group efficiencies of DMUs by accommodating the difference of their evaluates. It is widely known that the weights are not always uniquely determined. The mutual evaluation information which, the authors introduce, is calculated from the weights, is not always determined uniquely either. Here they propose a method for determining the weights uniquely by minimizing the square of the weights differences. The authors express the mutual evaluation information as a form of matrix which they call ‘Accommodation Efficiency Matrix.’ Furthermore, this paper discusses how to calculate the group efficiencies of DMUs by using maximum eigenvalue of this matrix. At the end, this paper shows an example which is the productivity analysis of Japanese electric power industries applying this method. [In Japanese.]