Time series clustering is wide spread problem in Data Stream Mining tasks and nowadays there are a lot of various approaches for solving such tasks that are based on different a priori assumptions. However, there are cases when well-known methods and algorithms for solving this task are inoperative in real applications. One of such tasks is short time series fuzzy clustering with unevenly distributed in time observations. The time series clustering of data set with missed observations is sufficiently close to this problem. The object of clustering is the sample in total and the observations are recorded by unevenly instants of time. Generated clusters are overlapped in such way that each processed sample can belong to several classes. At that it is assumed also, that all processed data are defined in the form of a fixed data set with unchanged size.
In the connections with that, it seems appropriate the spreading of the fuzzy clustering of short time series with unevenly distributed observations approach to the situation when the data are fed to the processing in online mode in the form of multivariate data stream in the context of Data Stream Mining.
In the paper the fuzzy clustering approach of multivariate short time series with unevenly distributed observations is considered. Such time series are fed to the processing in batch mode or sequentially on-line mode. In the first case we can use the matrix modification of fuzzy C-means method, and in second case we can use the matrix modification of neuro-fuzzy network by T. Kohonen, which is learned using the rule “Winner takes more”. Proposed fuzzy clustering algorithms are enough simple in computational implementation and can be used for solving of wide class of Data Stream Mining problems.
1. Liao, T.W. Clustering of time series data-A survey / T. W. Liao // Pattern Recognition. – 2005. – 38. – № 11. – P. 1857–1874.
2. Mitsa, T. Temporal Data Mining / T. Mitsa. -Boca Raton: CRC Press, 2010. – 395 p.
3. Aggarwal, C. C. Data Clustering. Algorithms, and Applications / C. C. Aggarwal, C. K. Reddy. – Boca Raton: CRC Press, 2014. – 621 p.
4. Aggarwal, C. C. Data Mining / C. C. Aggarwal. –N.Y.: Springer, 2015. – 734 p.
5. Möller-Levet, C.S. Fuzzy clustering of short time series with unevenly distributed sampling points / C.S. Möller-Levet, F. Klawonn, K.-H. Cho, O. Wolkenhauer // Lecture Notes in Computers Science. – Heidelberg: Splinger, 2003. – Vol. 2810. – P.330–340.
6. Cruz, L.P. Fuzzy clustering for incomplete short time series data / L. P. Cruz, S. M. Vieira, S. Vinga // Lecture Notes in Artificial Intelligence. – 9273. – Springer Int. Publishing Switzerland, 2015. – P. 353–359.
7. Bezdek, J. C. Pattern Recognition with Fuzzy Objective Function Algorithms / J. C. Bezdek. – N.Y.: Plenum Press. 1981. – 272 p.
8. Höppner, F. Fuzzy Clustering Analysis: Methods for Classification, Data Analysis, and Image Recognition / F. Höppner, F. Klawonn, R. Kruse, T. Runkler. – Chichester: John Wiley and Sons Ltd., 1999. – 289 p.
9. Bifet, A. Adaptive Stream Mining: Pattern Learning and Mining from Evolving Data Streams / A. Bifet. – IOS Press, 2010. – 224 р.
10. Bodyanskiy, Ye. Adaptive matrix fuzzy c-means clustering / Ye. Bodyanskiy, M. Skuratov, V. Volkova // Proc. 19th East-West Fuzzy-Colloquium. – Zittau-Görlitz: HS, 2012. – P.96–103.
11. Kohonen, T. Self-Organizing Maps / T. Kohonen // Berlin: Springer-Verlag. – 1995. – 362 p.
12. Haykin S. Neural Netwroks. A Comprehensive Foundation / S. Haykin. –Upper Saddle River: Prentice Hall, 1999. – 842 p.
13. Bodyanskiy, Ye. Matrix neuro-fuzzy self-organizing clustering network / Ye. Bodyanskiy, M. Skuratov, V. Volkova // Computer Science, Information Technology and Management Science. – 2011. – № 49. – P. 54–58.