A Novel Method of Data Stream Clustering Based on Wavelet Timing Series Tree Synopsis

Authors

Dongsheng Liu, College of Computer Science & Information Engineering, Zhejiang Gongshang University, Hangzhou 310018 P. R. China
Chonghuan Xu
Shujiang Fan

Author Country (or Countries)

Hangzhou 310018 P. R. China

Abstract

For the difficulty of obtaining cluster result fast and effectively under the limitations of bounded memory and time, this paper proposes a novel data stream clustering method based on wavelet timing series tree synopsis to solve the problem. The proposed method considers the attenuation characteristic of data stream, which combines the dynamic maintenance of wavelet coefficient and attenuation feature of wavelet coefficients of data stream, and can achieve approximate representation of data stream fragment information and dynamic maintenance of its synopsis structure. The proposed method employs wavelet timing series tree synopsis method to compress data stream fragment, then adopts two-phase density clustering algorithm to cluster. Detailed experiments show that the proposed method can get high compression quality, good space and time efficiency and good clustering results.

Suggested Reviewers

N/A

Recommended Citation

Liu, Dongsheng; Xu, Chonghuan; and Fan, Shujiang (2013) "A Novel Method of Data Stream Clustering Based on Wavelet Timing Series Tree Synopsis," Applied Mathematics & Information Sciences: Vol. 07: Iss. 3, Article 27.
Available at: https://digitalcommons.aaru.edu.jo/amis/vol07/iss3/27