基于主元的多元时间序列聚类分析方法研究

为提高多元时间序列聚类算法的效率,采用基于主元分析的多元时间序列聚类方法,将原始多元时间序列元素划分成一系列互不相关的簇,根据各簇的代表元素和剩余元素的主元素之间的扩展欧几里德范数对多元时间序列进行聚类分析.理论分析和实验结果表明该算法聚类质量结果和运行时间明显优于直接利用K-means法时的聚类结果.

Research on Analysis Method of Multivariate Time Series Clustering Based on Principal Component

To improve the efficiency of algorithms for multivariate time series(MTS) data clustering,a method of MTS clustering based on principal component analysis is proposed.A series of unrelated clusters are obtained by linear combination.According to extended Euclid norm the element and the remaining elements of the cluster between principal component,clustering analysis of MTS is done.The theoretical analysis and experimental results show that the clustering quality and operation time of proposed algorithm are better than that of direct K means clustering method obviously.