Improvement of models for determination of flexible factor type in data envelopment analysis.

Highlights • To put flexibility in the non-radial DEA models. • Presenting the flexible model as a mixed integer second order conic programming problem. • The new model gives higher efficiency score compared to the similar recent models in the literature. Abstract The role of some factors is not completely clear as an input or an output in many real applications of Data Envelopment Analysis (DEA). In other words, a factor can be used as an input by some Decision-Making Units (DMUs) while it may play an output role in other DMUs. This type of factors are called flexible factors. To improve the models for determining the type of flexible factors, two models were presented in this article in which flexible factors are added to be used in the best situation. In this way, factors in both output and input roles will optimize productivity with maximum efficiency without any change in the productivity level. The proposed models showed more consistency to achieve the desired goals in DEA as compared to similar models. Eventually, the proposed models were compared with existing models through a numerical example. [ABSTRACT FROM AUTHOR]