时间序列数据挖掘中的聚类研究综述  

Summary of Clustering Research in Time Series Data Mining

作　　者：李海林[1,2] 张丽萍[1] LI Hailin;ZHANG Liping(Department of Information Management and Information Systems,Huaqiao University Quanzhou Fujian 362021;Research Center for Applied Statistics and Big Data,Huaqiao University Xiamen Fujian 361021)

机构地区：[1]华侨大学信息管理与信息系统系,福建泉州362021 [2]华侨大学应用统计与大数据研究中心,福建厦门361021

出　　处：《电子科技大学学报》2022年第3期416-424,共9页Journal of University of Electronic Science and Technology of China

基　　金：国家自然科学基金面上项目(71771094)。

摘　　要：鉴于时间序列数据的高维性和复杂性给数据挖掘带来的困扰以及聚类分析在时间序列数据挖掘领域中的重要性,对目前该领域国内外相关时间序列数据聚类研究的状况进行综述。时间序列聚类总体上可分为整体时间序列聚类、子序列聚类和时间点聚类3种,分别从特征表示、相似性度量、聚类算法和簇原型等方面来研究,同时也结合了具体的应用分析。根据时间序列数据挖掘中聚类存在的主要问题,提出了部分未来值得关注和研究的内容和方向,以便更好地促进时间序列数据聚类分析的研究与发展。In view of the high dimensionality and complexity of time series data bringing trouble to data mining and the importance of clustering analysis in the field of time series data mining,this paper summarizes the research status of time series data clustering at home and abroad.Time series clustering can be divided into the whole-time-series clustering,the subsequence clustering,and it can be studied from the aspects of feature representation,similarity measurement,clustering algorithm and cluster prototype,as well as the specific applications analysis.According to the main problems existed in the time series clustering,this work proposes some contents and directions that are worthy of being researched in the future.All the work is to better promote the research and development of time series data clustering.

关 键 词：聚类分析 数据挖掘 高维性 时间序列 时间序列聚类 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]