Similarity Measure of Multivariate Time Series Based on Segmentation

Abstract

Similarity measure of time series is a fundamental problem in data mining tasks. However, most of the similarity methods are mainly for univariate time series, rather than multivariate time series. Among the existing approaches for multivariate time series, dynamic time warping can obtain high accuracy, but the calculation cost is expensive. To solve this challenging problem, a similarity measure of multivariate time series is proposed. We first segment multivariate time series and extract the mean value and span of each sequence as its feature. Then, a similarity measure based on dynamic time warping is proposed. Finally, extensive experiments on real-world data sets are executed. The experimental results indicate the proposed method can improve the efficiency while keeping the accuracy of similarity measure.