Research on Shape-Based Time Series Similarity Measure

Abstract

The representation and similarity measure of time series are the basis of time series research, and are quite important for improving the efficiency and accuracy of the time series data mining. In this paper, shape-based discrete symbolic representation and distance measure, which is used to measure the similarity between time series is presented. This method quantitatively represents the change of the shape of the time series. Compared with the approaches that exists similar, the present method is more intuitive and compact, and is not sensitive to offset translation, amplitude scaling, compress and stretch. That can reflect the degree of the dynamic change of the tendency and erase the influence of the noises, classify the patterns in more detail, which is favorable to improve the accuracy of the clustering, and multi-scale feature. The experimental results show that our approach has good effectiveness in clustering, which can satisfy the requirement of the shape-similarity of time series effectively under various analyzing frequency