Search

To efficiency measure in real-life applications of data envelopment analysis (DEA) model, the inputs and outputs data are sometimes imprecise, and we need to use the weight restrictions in order to incorporate management’s views. For this purpose, the present paper investigates the problems of existing approaches in this area, and proposes a comprehensive DEA model which enables us to estimate the relative efficiency scores of real-life systems. The proposed model contains different types of imprecise data and general form of weight restrictions. A new simulation-based genetic algorithm (GA) is developed to estimate the expected values of relative efficiencies with the comprehensive DEA model. It is shown that the proposed model and the solution approach solves the drawbacks of existing models, and gives more informative and reliable results. Some numerical examples are provided to illustrate the theoretical content of the paper and to show the effectiveness of the new approach.

In basic data envelopment analysis (DEA) models the status of input and output measures are exactly known and given. However, in certain cases, some factors could play either input or output roles that are referred to as flexible measures. The existing approaches to deal with flexible measures solve two models for each DMU to rank the DMUs, one to classify flexible measures, and another to estimate efficiency scores. Moreover, these approaches ignore the role of non-Archimedean epsilon in the performance evaluation process. In this paper, we develop a novel mixed binary linear DEA model to classify the flexible measures and find the best DMU. Moreover, we propose an approach to find a suitable epsilon value for our developed model and suggest a new algorithm to find and rank all efficient units. An interesting case study in Tehran Stock Exchange is provided to illustrate the potential application of our new approach.

This paper improves the integrated data envelopment analysis (DEA) model proposed by Asosheh et al. [1], for evaluating and ranking information technology (IT) projects in presence of both cardinal and ordinal data. They used balanced scorecard (BSC) for defining IT projects evaluation criteria and introduced a new DEA model to obtain most efficient IT project by considering imprecise data. This paper shows that their model randomly selects one efficient decision making unit (DMU) or IT project as the most efficient and the selection depends on the program or method is used for solving the problem. We develop their idea and propose a modified model to determine the most efficient DMU. In addition, this paper proposes a new algorithm for finding and ranking other efficient DMUs. The capability and usefulness of the proposed model is indicated by a numerical example.