An overview of clustering methods for geo-referenced time series: from one-way clustering to co- and tri-clustering. Issue 9 (1st September 2020)

• Record Type: Journal Article
• Authors: Wu, Xiaojing; Cheng, Changxiu; Zurita-Milla, Raul; Song, Changqing
• Is Part Of: International journal of geographical information science. Volume 34:Issue 9(2020)
• Page Start: 1822
• Page End: 1848
• Publication Date: 2020-09-01
• Subjects: Spatio-temporal pattern -- classification -- method selection -- clustering analysis -- data mining; Geography -- Data processing -- Periodicals; Information storage and retrieval systems -- Periodicals; Géomatique -- Périodiques; Systèmes d'information -- Périodiques; 910.285
• DOI: 10.1080/13658816.2020.1726922
• Languages: English
• ISSNs: 1365-8816

Abstract

ABSTRACT: Even though many studies have shown the usefulness of clustering for the exploration of spatio-temporal patterns, until now there is no systematic description of clustering methods for geo-referenced time series (GTS) classified as one-way clustering, co-clustering and tri-clustering methods. Moreover, the selection of a suitable clustering method for a given dataset and task remains to be a challenge. Therefore, we present an overview of existing clustering methods for GTS, using the aforementioned classification, and compare different methods to provide suggestions for the selection of appropriate methods. For this purpose, we define a taxonomy of clustering-related geographical questions and compare the clustering methods by using representative algorithms and a case study dataset. Our results indicate that tri-clustering methods are more powerful in exploring complex patterns at the cost of additional computational effort, whereas one-way clustering and co-clustering methods yield less complex patterns and require less running time. However, the selection of the most suitable method should depend on the data type, research questions, computational complexity, and the availability of the methods. Finally, the described classification can include novel clustering methods, thereby enabling the exploration of more complex spatio-temporal patterns.