Common weights for fully ranking decision making units by regression analysis

Existing methods for generating common weights in data envelopment analysis (DEA) are either very complicated or unable to produce a full ranking for decision making units (DMUs). This paper proposes a new methodology based on regression analysis to seek a common set of weights that are easy to estimate and can produce a full ranking for DMUs. The DEA efficiencies obtained with the most favorable weights to each DMU are treated as the target efficiencies of DMUs and are best fitted with the efficiencies determined by common weights. Two new nonlinear regression models are constructed to optimally estimate the common weights. Four numerical examples are examined using the developed new models to test their discrimination power and illustrate their potential applications in fully ranking DMUs. Comparisons with a similar compromise approach for generating common weights are also discussed.

Research highlights
► We propose a new methodology for seeking a common set of weights for fully ranking decision-making units (DMUs).
► Common weights are estimated with regression analysis techniques.
► Two nonlinear regression models are constructed.
► The discrimination power and potential applications of the proposed models are tested with four numerical examples.
► Comparisons with a compromise approach for generating common weights are discussed.