基于形态相似距离的时间序列相似度计算

摘要

Aiming at the shortcomings of the existing time series similarity distance measurements ,the morphology similarity distance (MSD) with the ability of shape discrimination was used to study the issue of time series similarity assessment .The similarity between financial time series was calculated based on the dynamic time warping distance ,the Euclidean distance and the morphology similarity distance after piecewise linear processing .Results of simulation indicate that the MSD is similar to the dy‐namic time warping distance on accuracy ,and it has the same computation speed as the Euclidean distance .%针对现有时间序列相似度距离计算方法的不足，应用具有一定形态区分能力的形态相似距离（morphology simila‐rity distance ，MSD）研究解决时间序列的相似度评估问题。完成金融时间序列的分段线性表示，分别基于动态时间弯曲距离、欧氏距离和0形态相似距离分析其相似度，仿真对比分析结果表明，形态相似距离的计算精度近似于动态时间弯曲距离，其计算速度近似于欧氏距离。