Research Of Methods And Their Application Of Determining The Weights Of Attributes In Fuzzy Comprehensive Evaluation

Comprehensive evaluation process and decision analysis are two inevitable work in daily life. A correct decision will make a great significant impact for not only person but nation, economy and society. With the development of society and the improvement of the complexity of things, people should take a lot factors into account when make decisions, which is multiple attribute decision problems. To determine the weight of each indicator is a critical aspect in multiple index comprehensive evaluation process. And weight distribution will have a direct impact on the final evaluation of the results. Therefore, this question is paid many attentions by scholars all around the world. Combined with present situation at home and abroad on the weights of attributes, we find it experienced from qualitative analysis to quantitative analysis, and then to the combination of qualitative analysis and quantitative analysis. This paper is based on the current weighted methods, then introduce different methods model, basic ideas and merits and demerits, and improve the entropy method, rough set method and the method of combination weighting model.Firstly, this paper introduces the mathematical model of fuzzy comprehensive evaluation to illustrate the key role of weight distribution of properties in this system. Then we summarize the main objective weighting model according to the original theory system and analyze the basic idea of the method, focus, advantages and disadvantages. We also summarize the application scope of various methods, the requirements on data type and whether the prior knowledge and measured data is necessary.Moreover, considering the limitation of the entropy method, that is, only the individual property of internal hidden rules are considered while ignoring the mutual influence between the indexes, we use conflict coefficient index, which fully reflect the relationship between indices, to modify weights of entropy method. And the factor analysis method and fuzzy clustering method are also be used to improve rough set method in the aspects of attribution reduction and continuous data processing. Finally, we further illustrate the method for the comprehensive weights combined with the application in daily life.