基于时间衰减和密度的任意簇数据流聚类

A data stream clustering algorithm based on time recession and arbitrary shape

数据挖掘的一个重要分支是数据流聚类技术。基于K均值算法的基础提出了CluTA算法。该算法在处理用K均值方法分类得到的结果时考虑时间衰减因素和相似簇的合并,达到用户对时间的要求并实现了任意形状簇聚类。理论分析和实验结果都表明算法具有可行性。

Data mining technology is an important branch of the data stream mine. This paper proposed a new algorithm named CluTA which based on kmeans algorithm. This algorithm consider time factor and merged similar sets when processed results of kmeans, it could realize users requirement of time limits and product arbitrary shape date set. Theoretic analysis and experimental results showed that CluTA is feasibility.