关键词: DEA模型, DEMATEL法, 创新效率, 超效率模型

Abstract: For a group of evaluation units with multiple inputs and multiple outputs, such as a group of industry-academia collaborative innovation projects to evaluate their relative effectiveness, the Data Envelopment Analysis (DEA) method is an effective and commonly used quantitative analysis method. However, in the optimization model constructed by the DEA method, because the feasible region does not fully consider the objectively important information among the indicators, it leads to the model overemphasizing the differences in the input and output characteristics of the evaluation unit, even led to an unreasonable phenomenon that the weight of a very important indicator is 0. An idea is proposed that combines DEMATEL method and data envelopment analysis method. That is to say, the index weight information obtained by the DEMATEL method is included in the constraints of the non-Archimedean limitless ε-CCR model and its super-efficiency model, and constructed a weight-constrained ε-CCR model and a weight-constrained super-efficiency ε-CCR model. Based on the above model, using the annual report data released by the US-Institute Collaborative Research Center of the National Science Foundation from 2014 to 2015, an empirical analysis of the relative efficiency of 71 university-led I/UCRC programs is conducted. The empirical results show that super-efficiency ε-CCR model with weight constraints is more effective than super-efficiency ε-CCR model without weight constraints, which can not only fully rank evaluation units, but also overcome the irrational evaluation result by super-efficiency ε-CCR model without weight constraints. These results help to further promote the theoretical innovation and application of DEA method, provide new methods for decision-making problems of multi-input and multi-output projects, and improve the scientific decision-making of related departments.

Key words: DEA model, DEMATEL method, innovation efficiency, super-efficiency model