带正则化项的时间序列聚类算法及其应用  

Clustering Method with Regularization for Time Series Data and Its Application

作　　者：陈南[1] 钟敏[1] 许伯熹[1] 

机构地区：[1]复旦大学数学科学学院,上海200433

出　　处：《复旦学报（自然科学版）》2012年第4期450-457,共8页Journal of Fudan University：Natural Science

摘　　要：为了更加准确地对时间序列数据进行聚类分析,运用了带正则化项的时间序列聚类方法,并实现了该聚类方法的算法.将该方法应用于云南地区水准形变数据的实际研究中,以寻找类之间状态转移与地震的关系.数值结果同时表明了带正则化项的时间序列聚类方法比标准的K-means方法更有效,更有优势.In order to have a better clustering result of the time series data,a new time series clustering method with regularization is introduced.Applying this method into studying the horizontal deformation data of Yunnan province,the relationship between the transition of the states and the eruption of the earthquakes could be figured out.The numerical results show that,comparing with the standard K-means method,this approach manifests its efficiency.

关 键 词：正则化 时间序列聚类方法 标准K-means方法 水准形变 

分 类 号：O29[理学—应用数学]