Cluster Analysis Approach for A Non-Homogeneous DMU’s Problem in DEA

44466

Data Envelopment Analysis (DEA) is a non-parametric method which is based on a linear programming. DEA aims at measuring relative efficiency of economic units employing similar inputs and producing the same kinds of outputs. Used Decision Making Unit (DMU) in DEA must use similar inputs/outputs and should operate under the same kind of environmental conditions. Since DEA is based on benchmarking against the best DMU, an existing variability in the external environmental conditions may affect dramatically the relative efficiency of the DMU’s. In this situation, DMU’s operating under the same kind of environmental conditions should be grouped into homogeneous classes according to environmental conditions having high variability. The better approach is to compare DMU’s in homogeneous class with each other. In this paper firstly, used DMU’s operating over large geographical area and in the same service industry, are grouped into homogeneous classes by K-means clustering algorithm. Then, DEA is applied for each separate homogeneous group and these results are compared with the results of DEA including all DMUs.

81

Data envelopment analysis , k , means clustering , efficiency

Ege Academic Review (EAR)

90

Ege Academic Review (EAR)