Weight restrictions in Data Envelopment Analysis: A comprehensive Genetic Algorithm based approach for incorporating value judgments

摘要

The basic DEA model experiences the weights flexibility problem which is resolved by the method of weight restrictions. The current research incorporating Decision Makers' (DMs) p into weight restrictions is subject to serious limitations such as lacking a framework for dual role factors and not incorporating organizational hierarchy in decision-making. The proposed Genetic Algorithm (GA) based approach for weight restrictions incorporates a dual role factor and organizational hierarchy in decision-making. The approach involves finding a set of weights which are at a minimum distance from all the DMs' p. The approach is flexible and is able to generate a common set of weights and Decision Making Unit (DMU) specific weight restrictions simultaneously. Results from model validation in a well-known automobile spare parts manufacturer in India indicate that the majority of suppliers perceived as highly efficient were actually found to be inefficient in the GA based weight restrictions model. A major contribution of this study is a robust approach to deal with multiple DMs and DEA weights flexibility problem. Another key highlight of the research is translating DMs p into a distance function. Using that as a fitness measure within the proposed Evolutionary Algorithms has been done for the first time in the presence of multiple DMs.