IMPROVEMENT INDICES KEEPING THE FEASIBILITY IN DATA ENVELOPMENT ANALYSIS

In this study, we propose a computational algorithm to determine the equations simultaneously, which define the efficient frontiers of the four types of DEA(Data Envelopment Analysis) models(the CRS, VRS, IRS and DRS models). By applying the algorithm, we calculate each efficiency score of all DMUs(Decision Making Units) and classify efficient DMUs into three types of RTS(Returns To Scale). Moreover, for each DMU which is not efficient in the CRS model, we present two kinds of improvements. One is for becoming an efficient unit in the CRS model. The other is for becoming an efficient unit in the CRS model over the PP(Production Possibility) set of the VRS model. For each type of improvements, we use an order matrix defined by the decision maker to calculate an improvement target contained in the PP sets of some models.