Optimality conditions and solution methodology for parameter selection in DEA (Q548462)

Summary: This paper focuses on developing optimality conditions and a solution methodology for generalised Data Envelopment Analysis (GDEA). The GDEA framework determines input and output parameters in DEA models of internal managerial performance of Decision Making Units (DMUs) based on an optimisation criterion involving an external non-managerial reward function on DMU performance. The resulting problem is a difficult non-convex integer optimisation model. In this paper, first-order optimality conditions are first derived, and then, these are utilised within a hybrid scheme of gradient-based search and a metaheuristic such as simulated annealing for efficient solution. Data from a financial application are used to illustrate the methodology.