The Application Of Fuzzy Clustering Method In The Classification Of Urban Comprehensive Competitiveness

"11th Five-Year Plan" period is a critical stage to build a well-off society. In order to enhance China's overall economic strength and to improve the people's standard of living, it's important strategy to focus on the development of large or core cities, which can give active effect to the areas around. Following this idea, we can improve the whole country rapidly and harmoniously. Meanwhile, "11th Five-Year Plan" concerning national economic and social development, stresses we should employ scientific development concept to guide overall, we should pay attention to improving the quality of economic growth and efficiency, and focus on resources saving and environmental protection. Consequently, to strengthen national comprehensive competitiveness, it's key point to appreciate and improve urban comprehensive competitiveness. It's significant to identify the cities' position to make appropriate policies.This article mainly discusses the classification of cities analyzed by fuzzy clustering method. Because of too many indicators, the boundary of different types are not clear, fuzzy clustering is a method of unsupervised classification, by the application of this method the result will be much more natural and realistic.Firstly, this paper studies research status of comprehensive competitiveness, and analyzes the meaning of this concept. Then four algorithms of fuzzy clustering have been compared, each has its advantage and disadvantages. There will be distortion of information when transitive closure is calculated in fuzzy clustering method based on equivalent matrix. The distortion is quiet obvious especially when large amount of data will be dealt with. FCM method requires prior knowledge of cluster prototypes and Category number. The efficiency of iterative algorithm varies greatly with different cluster prototypes, and sometimes only a partial optimal solution is obtained instead of a global one. After these two algorithms having been studied, a mixed fuzzy cluster model is put forward which can be described as follows: first, initial classification of cities is got through clustering method based on equivalence matrix, and when seeking equivalent matrix policymakers' inclination is taken into account, that is the index weight. Then suitable initial classification is chosen to be analyzed further by FCM method. In this way these two method can be compliment each other. In Section 5, related data of 15 sub-provincial cities and 9 cities of Wuhan Municipal Economy Circle are collected to do fuzzy clustering research. I expect the result can do some help to the location of Wuhan city. Empirical studies show that by employing the method in this paper, more stable results can be secured and the number of iteration can be largely reduced. Finally, the future research expected is discussed.