点击次数：

论文类型：期刊论文

发表时间：2009-12-01

发表刊物：1st International Symposium on Intelligent Informatics (ISII 2008)

收录刊物：SCIE、CPCI-S、Scopus

卷号：5

期号：12B

页面范围：4819-4824

ISSN号：1349-4198

关键字：Stock categories; Ant clustering; FCM clustering

摘要：Stock category is an important issue in stock analysis, Ant clustering algorithm, and Fuzzy C-Means (FCM) am two commonly used technologies for studying this problem.. However, there are some limits when just use each of them alone, Although the traditional ant clustering algorithm is capable of global searching and parallel computing, it also exist many problems including long time of clustering, poor convergence accuracy and so on. On the other hand, FCM is all effective clustering method, but it call not determine clustering number as well as clustering center by itself, which result will immerse into part optimal solution easily. So, we try to combine ant algorithm and FCM method to analyze the issue of stock categories. Ant algorithm, is used to determine the clustering number and clustering center, and clustering process can be dealt with FCM clustering algorithm. The results show that this new method call overcome the deficiencies of each single method and is more reliable.