基于时间序列与多部图的演化聚类动态分析

Morphing Cluster Dynamics Analysis Based on Time Series and Multipartite Graph

提出演化聚类动态分析法,使用图论理论,将时间序列数据集通过聚类分析表示为一个多部图。该图实现了对时间序列之间的群体级和个体级动态交互行为的捕捉和建模,不但能够追踪各个时间序列的来龙去脉,而且能对捕获的交互行为加以利用。

A new time series analysis algorithm named morphing cluster dynamics analysis is put forward. It reduces the time series set into amultipartite graph called system morphing graph. With this graph, the time series interactions on both levels are modeled and represented.