A novel clustering method on time series data

Time series is a very popular type of data which exists in many domains. Clustering time series data has a wide range of applications and has attracted researchers from a wide range of discipline. In this paper a novel algorithm for shape based time series clustering is proposed. It can reduce the size of data, improve the efficiency and not reduce the effects by using the principle of complex network. Firstly, one-nearest neighbor network is built based on the similarity of time series objects. In this step, triangle distance is used to measure the similarity. Of the neighbor network each node represents one time series object and each link denotes neighbor relationship between nodes. Secondly, the nodes with high degrees are chosen and used to cluster. In clustering process, dynamic time warping distance function and hierarchical clustering algorithm are applied. Thirdly, some experiments are executed on synthetic and real data. The results show that the proposed algorithm has good performance on efficiency and effectiveness. (C) 2011 Elsevier Ltd. All rights reserved.