多元时间序列因果关系分析研究综述

多元时间序列的因果关系分析是数据挖掘领域的研究热点. 时间序列数据包含着与时间动态有关的、未知的、有价值的信息, 因此若能挖掘出这些知识进而对时间序列未来趋势进行预测或干预, 具有重要的现实意义. 为此, 本文综述了多元时间序列因果关系分析的研究进展、应用与展望. 首先, 本文归纳了主要的因果分析方法, 包括Granger因果关系分析、基于信息理论的因果分析和基于状态空间的因果分析; 然后, 总结了不同方法的优缺点、适用范围和发展方向, 并概述了其在不同领域的典型应用; 最后, 讨论了多元时间序列因果分析方法待解决的问题和未来研究趋势.

Survey on Causality Analysis of Multivariate Time Series

The causality analysis of multivariate time series is a research hotspot in data mining.Time series data contains unknown,valuable information related to temporal dynamics.Therefore,it is of great practical significance to be able to mine these knowledge and then predict or intervene the future trend of time series.For this reason,this paper reviews the research progress,application and prospects of causality analysis of multivariate time series.Firstly,this paper summarizes the main causality analysis methods,including Granger causality analysis,causality analysis based on information theory and causality analysis based on state space.Then,we summarize the advantages and disadvantages,scope of application and development directions of different methods,and outline their typical applications in different fields.Finally,the problems to be solved and future research trends of the causality analysis methods of multivariate time series are discussed.