Abstract
Recently, multi-objective optimization are getting more and more important to have better considerable result in actual problems. And multi-objective optimization is used in various fields. However, when we have more than three objective functions, it is not easy to grasp its situations. In that sense, visualization of multi-objective optimization becomes quite important issue. When we use DEA, we can have optimum weights and efficiency values of objective function for each DMU, but it is still difficult to see its relationships visually for them. In post study, we have proposed DEA mapping. In DEA mapping we use Lagrange multiplier of extended CCR model as key information of mapping. Sum of product of Lagrange multiplier and data becomes a point on frontier that the unit should aim in original space. Extended CCR model remove the upper limit of the efficiency values of each DMU, and it can acquire the relative merits for DMU on the frontier. We make reduction of multidimensional data using PCA as a starting position of each DMU. We assumed projection relationship of each DMU to frontier still remains in the reduction space. Therefore we can calculate projection point by using position data and Lagrange multiplier, Distance between DMU's original position and projection point makes contradiction. In this paper, we examine DEA mappings' character and indication make of how to be visualize relationships easy for designers. This method can figure out relationships of each DMU.
