Selecting Appropriate Variables for DEA Using Genetic Algorithm (GA) Search Procedure

Abstract

Data envelopment analysis (DEA) is one of the most powerful non-parametric methods to assess the relative efficiency of each Decision making units (DMU’s). Its simplicity in computation and mathematical programming technique attracted many researchers and at the same time DEA is more sensitive to variables considered. It uses multiple inputs and outputs for efficiency analysis but does not provide any guidelines in choosing variables and hence researchers selected their own number of input and output variables using several methods. Usage of all the variables in DEA is not sensible, since irrelevant variables will reduce the efficiency power. Therefore, selection of appropriate or best set of variables for input and output is needed but it’s one of the crucial tasks in DEA. In the present paper, a new approach of selecting appropriate set of variables using genetic algorithm are discussed and applied to Indian banking sector.