基于滑动窗口及局部特征的时间序列符号化方法

摘要

This paper put forward a symbolic method for time series which based on the sliding window and the local features. Firstly, this method divided time series according to the sliding window method, and used multiple slopes to show the each subsection of time series. Then it used K-means clustering algorithm to clustering the slope representation of subsection, and realized the symbolization of time series. Experiments show that the method is effective and accurate.%针对时间序列的数据挖掘将时间序列数据转换为离散的符号序列,提出了一种基于滑动窗口及局部特征的时间序列符号化方法.该方法采用了滑动窗口的方法将时间序列分割,每个分段采用多个斜率表示,最后采用K-均值聚类算法对斜率表示的分段进行聚类,实现时间序列的符号化.实验证明了该方法的有效性与准确性.