Abstract: In the process of comprehensive evaluation, decision-makers not only need to obtain the evaluation results of an evaluated object, but also need to know the reasons leading to such results and the feasible optimization paths. The data envelopment analysis (DEA) method and fuzzy comprehensive evaluation (FCE) method are two very important but mutual independent evaluation methods. If the correlation between the two methods can be found, it not only can enhance the advantages of the two methods each other, but also may provide new perspectives and tools for the solution of the above problems. In addition, it will be a very difficult task to find the improved information of synthesized fuzzy indicators by applying DEA method.
To address these issues, this study first constructs a fuzzy multi-level possible sets based on the existing or obtained FCE information, and provides the definition of fuzzy multi-level projection and its measurement method. Then, these methods are applied to analyze the cause of FCE results, and find the improved strategies of multi-level fuzzy objects. Finally, the proposed method is illustrated with an empirical study of the tourist satisfaction of 14 tourism provinces and regions in China.
Based on the results of fuzzy evaluation, this paper further analyzes the existing problems and improvement direction of the tourist satisfaction of 14 tourism provinces in China. The results of the empirical analysis show that the decision-makers can obtain evaluation information and correlation relationships at different levels within the fuzzy object by applying the method in this paper. These results not only reveal the improvement direction and feasible scale of the evaluation object, but also find a feasible way to integrated DEA and FCE.
Due to the lack of previous work on applying DEA ideas to improve FCE methods, this study is only a preliminary attempt, and still needs to be further improved and deepened. For example, the fuzzy operator selected in this paper is the weighted average operator, which is the most commonly used. However, the FCE method has other operators such as the main factor-determining type or the main factor-emphasizing type. Furthermore, more methods can also be used to construct the possible set of evaluation results. Further research in these areas will play a positive role in improving the application ability of multi-level FCE method and enriching FCE techniques.

Key words: multi-objective decision-making, data envelopment analysis, fuzzy comprehensive assessment, evaluation object, fuzzy matrix

摘要： 数据包络分析方法与模糊综合评判方法是两个十分重要但又相互独立的评价方法,如果能够找到两种方法的关联关系,进而实现两种方法的优势互补将是一项值得探讨的工作。另外,模糊指标合成后,要想应用数据包络分析方法找到更微观指标的改进信息也非常困难。为解决这些问题,本文首先基于模糊综合评判的已有信息,构建多层次模糊评价结果的可能集,并提出多层次模糊投影的定义及相应的计算公式。在此基础上,给出了测度模糊事件存在不足的定量方法。最后,以中国14个旅游省区的游客满意度为例进行了实证分析。

关键词: 多目标决策, 数据包络分析, 模糊综合评判, 评价对象, 模糊矩阵