基于时间序列的相似子模式发现算法

基于时间序列的数据挖掘时,一般需要对时间序列离散化,再聚类成不同的子模式。已有的方法常忽略时间序列本身的位置和整体特征,并且计算量大。针对其不足,文中提出一种检索时间序列分段关键点的算法,以关键点为边界分段,使用形态距离测度和快速剪除的算法,高效简便地检索出相似子模式。

An Algorithm of Finding Similar Subpattern Based on Time Series

Based on the data mining of the time series,generally discretizes time series,then clusters different sub-pattern.Some methods constantly ignore time series of position and dynamic attribute furthermore calculation large.For its shortage,presents searching time series key point algorithm,with the key point for boundary,using the appearance distance method,measures each cent a subsequence,mining latency information from the new angle of view.