A Weighted DTW Approach for Similarity Matching over Uncertain Time Series

Abstract

To measure uncertain time series similarity effectively and efficiently, in this paper, we propose a weighted DTW distance-based approach for uncertain time series with the expected distance. We introduce a weight function to assign weights to a reference point and a testing point. With this function and the WDTW, the accuracy of calculating uncertain time series similarity can be improved. Also, to reduce the storage space and time-consuming, we extend the lower bound function LB_Keogh for DTW into ULB_Keogh for our approach.