作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2007, Vol. 33 ›› Issue (12): 102-104. doi: 10.3969/j.issn.1000-3428.2007.12.036

• 软件技术与数据库 • 上一篇    下一篇

基于滑动窗口的多变量时间序列异常数据的挖掘

翁小清1, 2,沈钧毅1   

  1. (1. 西安交通大学计算机软件与理论研究所,西安 710049;2. 河北经贸大学计算机中心,石家庄 050061)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-06-20 发布日期:2007-06-20

Outlier Mining for Multivariate Time Series Based on Sliding Window

WENG Xiaoqing1, 2, SHEN Junyi1   

  1. (1. Institute of Computer Software, Xi’an Jiaotong University, Xi’an 710049; 2. Computer Center, Hebei University of Economics and Trade, Shijiazhuang 050061)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-06-20 Published:2007-06-20

摘要: 与其它多变量时间序列(MTS)子序列显著不同的子序列,称为异常子序列(含异常数据)。该文提出了一种基于滑动窗口的MTS异常子序列的挖掘算法,使用扩展的Frobenius 范数来计算两个MTS子序列之间相似性,使用两阶段顺序查询来进行K-近邻查找,将不可能成为候选异常子序列的MTS子序列剪去,对上海证券交易所股票交易情况MTS数据集进行了异常子序列(含异常数据)挖掘,结果表明了算法的有效性。

关键词: 多变量时间序列, 滑动窗口, 局部稀疏系数, 扩展的Frobenius范数, 异常数据挖掘

Abstract: Multivariate time series (MTS) subsequences, which differ significantly from the remaining MTS subsequences, are referred to as outlier subsequences. Tthe mining method for MTS outlier subsequences based on sliding window is proposed. An extended Frobenius norm is used to compare the similarity between MTS subsequences, K-NN searches are performed by using two-phase sequential scan, and MTS subsequences which are not possible outlier candidates are pruned which reduce the number of computations and comparisons. The MTS datasets of stock market is used for outlier mining, the results show the effectiveness of the algorithm.

Key words: Multivariate time series, Sliding window, Local sparsity coefficient, Extended frobenius norm, Outlier mining

中图分类号: