Robust data envelopment analysis with variable budgeted uncertainty

We present in this paper a new robust data envelopmentanalysis (DEA) model where the uncertain parameters are inform of the variable budgeted uncertainty, an extension ofthe budgeted uncertainty introduced by Bertsimas and Sim.Using variable budgeted uncertainty we aim to provide thesame probabilistic guarantee as the budgeted uncertaintyand, simultaneously, to obtain robust solutions that are lessconservative compared to Bertsimas and Sim’s uncertainty.The new presented robust DEA model is a mixed-integerprogramming problem and we provide the mathematicaldetails of the proposed model that are based on thedualization technique often used in robust linearprogramming. Finally, we present a computationalcomparison of the traditional robust DEA models that aredeveloped based on the budgeted uncertainty and the novelrobust DEA model. The experiments reveal that using theproposed robust DEA, the price of robustness reducescompared to traditional robust DEA models by an averagefactor of ۳۰%.

Keywords:

Data envelopment analysis , Robust optimization , variablebudgeted uncertainty , price of robustness