A feasible interval for weights in data envelopment analysis

In data envelopment analysis (DEA), sometimes it is required to impose restriction on weights based on the judgement of the decision maker (DM). These additional constraints may result in infeasibility of the model. Small alteration of the bounds will ensure the feasibility of the model. This alteration must be minimized in order to retain the preferences of DM. By using goal programming (GP) technique and exploiting big-M method, a set of weights are obtained for which the corresponding problem will be always feasible. It has been shown that the sum of deviations of these weights from managerial willingness is minimum. An empirical study using real data shows the robutness of the algorithm.