Analyzing a Hybrid Approach of PCA-DEA in Two Different Modes

Abstract

The present work compared input- and output-based integrated principal component analysis-data envelopment analysis (PCA-DEA) approaches. The approach minimizing the number of decision-making units (DMUs) identified as efficient would be the superior one (as it facilitates DMU ranking). This approach would somewhat handle a major drawback of DEA– i.e., the emergence of an excessively high number of DMUs. The input and output-based approaches were independently implemented in MATLAB and were compared to identify the superior one. A number of numerical examples were carried out to demonstrate the performance of the superior approach. The results show that the second approach (the output-based approach) is superior to the first approach (the input-based approach). Therefore, it is better to divide the outputs by the inputs to create the PCA-DEA indices. In order to achieve better results in this way, this point (of dividing the outputs by the inputs) is not specific to this research alone and can be used in other research (in case study research).