A Research And Application On The Fuzzy C-means Based On Rough Set

The technology of data mining is a cross subject developing rapidly domestically and overseas, It is a process that to deal with and extract unknown information and knowledge from large databases by using the means of database, statistics, artificial intelligence, and machine learning. It has been applied extensively in the fields of agriculture, finance, insurance, national defence and so on.Clustering is used to find out the objects that are resemble each other and compose different groups, cluster analysis is an important job in data mining. there are many ways to classify the dataset, the partitioning methods ,hierarchical methods and grid-based methods have great recognition. fuzzy cluster methods applied the theory of fuzzy mathematics in cluster analysis, can do with the data classification that has a fuzzy edge. It is more meaningful than previous hard classification.However, because the variety of the database structure, from now on, there is no one algorithm applied all cluster analysis. in the applications, a lot of cluster analysis tasks need a certain algorithm. Though the traditional fuzzy cluster algorithms can handle many problems that have fuzzy edges, there are many disadvantages also.By comparison and researches on the fuzzy cluster algorithms, This article aiming at the fuzzy clustering contributed in such aspects:firstly, introduce the theory basis of data mining and clustering analysis, emphasize on the fuzzy cluster algorithms' analysis and comparison, especially the comparison of FCM algorithm and NFWFCA algorithm.secondly, analyze the problems in the clustering analysis, attributes redundancy, the disproportion of attributes' weights and isolated points' sensitivity and so on. these problems lead to the decrease of corrected ratio and running efficiency.thirdly, for resoling these problems efficiently, this article brings the rough set theory into fuzzy cluster, by using the methods of attributes contracted in the rough set theory to improve the FCM algorithm, the improved algorithm had been proved a high precise ratio.finally, applied the improved algorithm into the information research system, by reclustering the search results, It lead to a high efficiency information research system.