时间序列的相似性测度

Measures of Time Series Similarity

时间序列(time series)是指按时间顺序排列的观测值集合,在生物信息学研究领域中,DNA序列和基因表达数据都可以视为时间序列数据。时间序列分析中很重要的环节就是刻划两个时间序列或者时间子序列的相似性,用于序列比对等。时间序列的相似性测度是时间序列研究中的基础和重点,直接影响查询、聚类等后续计算的效率和精度,在高通量基因芯片数据分析、基因网络构建等研究中,具有重要的应用,目前已引起了众多研究人员的关注,在欧氏距离的基础上进行了大量的研究,本文综述了基于欧式距离和时间弯曲的时间序列相似性测度及其相关领域的研究进展,可作为进一步研究的参考。

The time series refer to the observation value sets ordering with time sequence such as DNA sequences and genes expression data in bioinformatics. The important research is describing the similarity between two time series or time subsequences used in sequence aligning,for instance. The similarity measures are essential for time series research which will inference the efficiency and accuracy of query, clustering and so on, especially, for high - throughout gene arrays analysis and gene networks building. More scientists have paid more attention on it. This paper reviewed the advances of time series similarity measures based on the Euclidean and time warping distances for reference.