Improving discrimination power based on reducing dispersion of weights in data envelopment analysis

The main drawbacks that arise for data envelopment analysis (DEA) are: lack of discriminationpower amongst efficient decision making units (DMUs) and scattering input-output weights. In theDEA, sometimes the mismatch of the input or output weights in the decision-making units (DMUs)under consideration leads to assigning higher weight to variables with the less significance and/or thelower or zero weight to the variables with high significance. Accordingly, most DEA models introducemore than one efficient DMU in evaluating the relative efficiency of decision-making units. The presentpaper is conducted to overcome these inabilities. In this trends, we present a novel DEA model basedon minimizing the sum of absolute deviations of all input-output weights from each other. The proposedmodel provides to enhance the discrimination power and adjusts the balance dispersion of input-outputweights. Finally, well-known numerical experiments are considered to demonstrate the efficiency andvalidation of the suggested model.