Improved DEA model by quantifying the extreme degree of weight schemes

Abstract Data Envelopment Analysis (DEA) is a kind of nonparametric methodology used to evaluate different DMUs. DEA allows different DMUs to choose a set of weights that can maximize their efficiency score. Unfortunately, the flexibility in choosing weights is not only an advantage but also a source of disadvantages of DEA. To overcome the shortfalls of the traditional DEA model, a new DEA model based on quantifying the extreme degree of weights is proposed in this paper. The new model uses a pair of parameters and a set of inequalities to limit the selection range of weights. In order to get evaluation results independent of parameters, a novel method based on calculating the area under curve is proposed. Furthermore, the reasonableness of the new model is illustrated by comparing the results of a numerical example between the proposed model and the popular models. Finally, some conclusive remarks and suggestions are proposed for future research.