| Summary: | 博士 === 義守大學 === 工業管理學系 === 105 === Data envelopment analysis (DEA) is used to evaluate the efficiency of decision-making units (DMUs), but sometimes the number of DMUs or the number of inputs and outputs may vary. Therefore, sensitivity analysis is often performed to investigate how these variations affect DEA outcomes. Sensitivity analysis allows decision-makers to re-examine all the important decision variables. When the data input to a DEA model are uncertain, sensitivity analysis becomes crucial.
Using sensitivity analysis, this study investigated how changes in the input or output of each DMU affected the assessment results if the other conditions were unchanged. First, the stability radius was clearly defined. Second, this study proposed innovative methods to assess the stability radii of DMUs in the DEA/Charnes–Cooper–Rhodes and DEA/common weights models.
Mathematical proof of the proposed methods is provided, and real data from 16 firms in the Taiwanese textile industry are used as a case study to explain how the methods can be applied in real situations. The study revealed that sensitivity analysis of each DMU can be performed by assessing the stability radius of the DMU. Additionally, the stability radius can be used as a criterion to re-rank multiple efficient DMUs and hence increase the subtlety of an efficiency evaluation.
|
|---|
