Computer Engineering and Applications ›› 2012, Vol. 48 ›› Issue (23): 111-114.

Previous Articles     Next Articles

Analysis on similarity of multivariate time series based on Eros

GUO Xiaofang1, LI Feng2   

  1. 1.School of Computer Science and Engineering, Jiangsu University of Science and Technology, Zhenjiang, Jiangsu 212003, China
    2.School of Electronics and Information, Jiangsu University of Science and Technology, Zhenjiang, Jiangsu 212003, China
  • Online:2012-08-11 Published:2012-08-21

基于Eros的多元时间序列相似度分析

郭小芳1,李  锋2   

  1. 1.江苏科技大学 计算机科学与工程学院,江苏 镇江 212003
    2.江苏科技大学 电子信息学院,江苏 镇江 212003

Abstract: In order to improve the similarity search efficiency of Multivariate Time Series(MTS), Principal Component Analysis(PCA) method of Extended Frobenius norm(Eros) is used. Principal component similarity factors, which are composed of the main element and the eigen values, are used for the comparison of multivariate time series similarity matrix. In order to verify the validity of this method, experiments on real data set or synthesis experiment data set are carried out separately, the results show that, the proposed method has certain superiority than the traditional Euclidean Distance(ED), Dynamic Time Warping(DTW) similarity measure method.

Key words: multivariate time series, principal component analysis, extended Frobenius norm, recall-precision

摘要: 为提高多元时间序列相似性度量的效率,采用扩展Frobenius范数(Eros)的主元分析(PCA)方法,通过主元和本征值构造主元相似因子,用于比较多元时间序列矩阵之间的相似性。为了验证这种方法的有效性,针对三组数据(两个真实数据,一个合成数据)进行了实验。结果表明,该方法相对于以往的欧几里德距离(ED),动态时间弯曲(DTW)相似性度量方法具有一定的优越性。

关键词: 多元时间序列, 主元分析, 扩展Frobenius范数(Eros), 查全率-查准率