Integration of data envelopment analysis with decision maker preference for supplier selection

Data Envelopment Analysis (DEA) is a well-known method for selecting suppliers. Although Data Envelopment Analysis can compare suppliers in terms of their efficiencies, it cannot capture suppliers’ effectiveness and decision maker’s preference over criteria in the suppliers’ scores. We introduce a function called the Preference Function to capture the decision maker’s preference over different criteria. After applying Preference Functions on criteria’s data, we run Data Envelopment Analysis. Two different methods are proposed to derive the Preference Functions: direct and feedback-based (indirect). In the direct method, the decision-maker directly tailors Preference Functions to her preferences toward different criteria. In the feedback-based method, Preference Functions are formed via an optimization scheme. In this method, a mathematical model is solved, and then the result is used to iteratively build Preference Functions. We deploy the Particle Swarm Optimization technique to solve this problem. We illustrate both direct and indirect methods through solving a supplier selection example and compare it with plain Data Envelopment Analysis. Finally, through extensive numerical analysis, we show that Particle Swarm Optimization effectively solves the problem in the indirect method.