Dynamic Matrix Clustering Method Based on Time Series
AUTHORS
Lu Liu,Harbin Institute of Petroleum, Harbin, Heilongjiang 150027, China
ABSTRACT
Time series event clustering is the basis of research event classification and mining analysis. Existing clustering methods mostly directly cluster continuous events with time attributes and complex structures, without considering the conversion of clustering objects, and the accuracy of clustering results is low, and the efficiency is poor. To solve these problems, a dynamic matrix clustering method for time series events is proposed. Construct an event neighbor evaluation system, and construct a candidate set through the backward difference calculation strategy of neighbor scores. This paper proposes a method for selecting candidate sets of diverse sequences based on combinatorial optimization, and quickly obtains the optimal solution of the diverse sequence RDS from the candidate sets. Finally, the distance matrix between RDS and the data set is dynamically constructed, and a matrix clustering method based on K-means is proposed to realize the effective division of the categories of time series events.
KEYWORDS
Time series, Dynamic matrix clustering, Combinatorial optimization, K-means
REFERENCES
[1] Li Hailin and Guo Chonghui, “Research review of feature representation and similarity measurement in time series data mining,” Computer Application Research, vol.30, no.5, pp.1285-1291, (2013)
[2] Yang Yiming, Pan Rong, Pan Jialin, etc., “Algorithm comparison of time series classification,” Chinese Journal of Computers, vol.30, no.8, pp. 1259-1266, (2007)
[3] Cong Si'an and Wang Xingxing, “Overview of K-means algorithm research,” Electronic Technology and Software Engineering, no.17, pp.155-156, (2018)
[4] Chen Chaowei and Chang Dongxia, “Automatic clustering algorithm based on density difference,” Journal of Software, vol.29, no.4, pp.935-944, (2018)
[5] Shu K., Sliva A., Wang S, et al., “Fake news detection on social media: A data mining perspective,” ACM SIGKDD Explorations Newsletter, vol.19, no.1, pp. 22-36, (2017)
[6] Euán C., Ombao H., and Ortega J., “The hierarchical spectral merger algorithm: a new time series clustering procedure,” Journal of Classification, vol.35, no.1, pp.71-99, (2018)
[7] Azencott R., Muravina V., Hekmati R., et al., “Automatic clustering in large sets of time series,” Contributions to Partial Differential Equations and Applications. Springer, Cham, pp.65-75, (2019)
[8] Zhang D. Y., Zhou L. H., Wu X. Y., and Zhao L. H., “Data stream clustering based on grid coupling,” Journal of Software, vol.30, no.3, pp.667-683, (2019)
[9] Zheng J. W., Li Z. R., Wang W. L., and Chen W. J., “Clustering with joint Laplacian regularization and adaptive feature learning,” Journal of Software, vol.30, no.3, pp.3846-3861, (2019)
[10] Zakaria J., Mueen A., and Keogh E., “Clustering time series using unsupervised-shape lets,” IEEE 12th International Conference on Data Mining, pp.785-794, (2012)
[11] Madiraju N.S., Sadat S.M., Fisher D., and et al., “Deep temporal clustering: Fully unsupervised learning of time-domain features,” arXiv preprint arXiv:1802.01059, (2018)