Науковий вісник НГУ

Information technology of the multivariate time series fuzzy clustering on the example of the Samara river hydrochemical monitoring

• 
• 

User Rating:  / 0

PoorBest 

Details

Category: Information technologies, systems analysis and administration

Last Updated on 13 November 2014

Published on 13 November 2014

Hits: 3996

Authors:

O.G. Baibuz, Dr. Sci. (Tech.), Professor, Oles Honchar Dnipropetrovsk National University, Head of the Department of Software Design, Dnipropetrovsk, Ukraine

M.G. Sidorova, Oles Honchar Dnipropetrovsk National University, Postgraduate Student, Dnipropetrovsk, Ukraine

Abstract:

Purpose. Development of the methods for filling the information technology of fuzzy clustering in the case of multivariate time series.

Methodology. This paper presents a technique of cluster analysis of multivariate time series as a computational schemes based on the one-dimensional time series clustering, aggregating results into a similarity matrix and determination the result fuzzy partition.

Findings. Computational schemes of methods: agglomerative hierarchical, K-means, Forel, graph method of the shortest non-closed path have been adapted to the time series clustering and included in the core of the proposed information technology. Their quality has been assessed by different quality criteria. The practical implementation with the analysis of the results has been applied to the data of hydrochemical monitoring of technologically-laden area.

Originality. A new metric for comparing time series which takes into account both the nature of the compared series and the closeness of their values has been proposed, that can improve the quality of clustering. The information technology of multivariate time series clustering based on an ensemble approach and fuzzy logic has been proposed.

Practical value. On the basis of the proposed technology and developed software cluster analysis of data hydrochemical monitoring of surface waters of the West Donbass region (r. Samara) has been held. It has allowed to identify groups of control points, which characterized by similar physical and chemical composition of water on the investigated components for proper environmental planning and management of water quality of the river.

References:

1. Wang, X., Smith, K., Hyndman, R. and Alahakoon, D. (2001), “A scalable method for time series clustering”, Tech. Report Department of Econometrics and Business Statistics at Monash University, Melbourne, Australia.

2. Паршутин С.В. Кластеризация временных рядов с применением карт самоорганизации: сборник научных трудов / С.В. Паршутин // Интегрированные модели и мягкие вычисления в искусственном интеллекте. – Коломна. – 2007. – C. 465–472.

Parshutin, S.V. (2007), “Time series clustering with application of Self-Organizing Maps”, Int. Conf. Proc. “Integrated Models and Soft Computing in Artificial Intelligence”, Kolomna, pp. 465–472.

3. Iglesias, F. and Kastner, W., (2013), “Analysis of similarity measures in times series clustering for the discovery of building energy patterns”, Energies, Vol. 6, pp. 579–597.

4. Alcock, R.J. and Manolopoulos, Y., (1999), “Time-Series Similarity Queries Employing a Feature-Based Approach.”, 7th Hellenic Conference on Informatics. August 27–29, Ioannina, Greece.

5. Liao,T.W.,(2005), “Clustering of time series data – survey”, Pattern Recognition, Vol. 38,pp. 1857–1874.

6. Rani, S., Sikka, G., (2012), “Recent Techniques of Clustering of Time Series Data: A Survey”, International Journal of Computer Applications, Vol. 32, no.15, pp. 1–9.

7. Гусарова Л. Проверка обоснованности кластерного решения / Л. Гусарова, И. Яцкив // Reliability and statistics in transportation and communication (RelStat). – 2004. –Т. 5. – №2. – С. 49–56.

Gusarova, L. and Yatskіv, I., (2003), “Checking of the cluster solution validity”, Proc. of Int. Conf. RelStat, Vol.5, no. 2, pp. 49–56.

8. Вятченин Д.А. Нечеткие методы автоматической классификации: монография / Вятченин Д.А. – Минск:УП „Технопринт“, 2004. – 219с.

Vyatchenin, D.A., (2004), Nechetkie metody avtomaticheskoy klassifikatsii [Fuzzy Methods of Automatic Classification], Monograph, Tehnoprint, Minsk, Belarus.

9. Яцкив И. Методы определения количества кластеров при классификации без обучения/И. Яцкив, Л. Гусарова // Transport and Telecommunication. – 2003. – T.4. –  №1. – C. 23–28.

Yatskіv, I. and Gusarova, L., (2003), “Methods for determining the number of clusters in unsupervised classification”, Transport and Telecommunication, Vol.4, no. 1, pp. 23–28.

2014_5_baibuz
2014-11-10  229.46 KB  918