A Method for Fast Clustering in Time Series Database based on SPDTW
Yong Wang · Journal of Guangxi University for Nationalities · 2010
In this paper we introduce a novel dimensionality reduction technique for Time series which we call Stratified Piecewise Aggregate Approximation(SPAA).And furthermore,we propose a new method allow fast Clustering in large Time Series database,which we call Stratified Piecewise Dynamic Time Warping.And then,we theoretically and empirically compare it to the classical DTW and demonstrate its superiority.Using SPDTW is up to several orders of magnitude than classical DTW,hardly without accuracy loss.