Data envelopment analysis in the presence of correlated evaluation variables

Abstract

Data Envelopment Analysis (DEA) is a technique for evaluating homogeneous Decision-Making Units (DMUs) that consume similar inputs to produce similar outputs. An essential principle in this method is to identify inputs and outputs; the identified inputs (outputs) must be independent of each other. However, in the real world, there are situations where there is a correlation between two or more inputs (outputs), and then one of them should be considered in the performance evaluation. This issue can cause problems in practice. The main question, in this case, will be that" Which of these two or more correlated variables should be considered in evaluating DMUs?". In this paper, a method for determining an essential variable using a DEA model is presented. In this way, the basic models of DEA have been integrated with the 0-1 programming to achieve the above objective. The proposed method is then improved by using Centralized Data Envelopment Analysis (CDEA) model, followed by refining the performance evaluation variables. At last, the application of the proposed method has been verified for different examples. Results show that the proposed method selects the appropriate variable from among the correlated variables. Also, improving the method using a centralized approach leads to the selection of a variable that increases the total efficiency. The application and implementation of the proposed method is simple and does not have computational complexity. It also does not need experts’ judgment, so it is a cost-effective way © 2021 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved