ST Sequence Miner: visualization and mining of spatio-temporal event sequences
- PDF / 1,174,023 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 82 Downloads / 208 Views
ORIGINAL ARTICLE
ST Sequence Miner: visualization and mining of spatio-temporal event sequences Baran Koseoglu1
· Erdem Kaya1
· Selim Balcisoy1
· Burcin Bozkaya1,2,3
© Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract As a promising field of research, event sequence analysis seems to assist in facilitating clear reasoning behind human decisions by mining reality behind the sequential actions. Mining frequent patterns from event sequences has proved to be promising in extracting actionable insights, which plays an important role in many application domains. Much of the related work challenges the problem solely from the temporal perspective omitting the information that could be gained from the spatial part. This could be in part due to the fact that analysis of event sequences with references to both time and space is attributed as a challenging task due to the additional variance in the data introduced by the spatial aspect. We propose a visual analytics approach that incorporates spatio-temporal pattern extraction leveraging an extended sequential pattern mining algorithm and a pattern discovery guidance mechanism operating on geographic query and selection capabilities. As an implementation of our approach, we introduce a visual analytics tool, namely ST Sequence Miner, enabling event pattern exploration in time-location space. We evaluate our approach over a credit card transaction dataset by adopting case study methodology. Our study unveils that patterns mined from event sequences can better explain possible relationships with proper visualization of time-location data. Keywords Sequence mining · Event sequences · Spatio-temporal data · Information visualization · Visual analytics
1 Introduction Recent availability of and the tremendous increase in volume of accumulated data in business realms led to growing interest in research on event sequence analysis. More importantly, extraction of causality behind the investigated phenomenon is attributed as crucial since such efforts can lead to actionable insights. As a promising field of research, event sequence analysis has attracted much attention from the literature as it can reveal more reality than mere relationships [29]. Exploration of event sequences by means of visual analytics systems has been investigated with a variety of visualization techniques [21]. Recent studies further
B
Baran Koseoglu [email protected]
1
Behavioral Analytics and Visualization Lab, Sabanci University, 34956 Orhanli, Tuzla, Istanbul, Turkey
2
Sabanci Business School, Sabanci University, 34956 Orhanli, Tuzla, Istanbul, Turkey
3
New College of Florida, 5800 Bay Shore Road, Sarasota, FL 34243, USA
integrate sequential pattern mining (SPM) or sequence clustering techniques to facilitate the extraction of frequent event sequences from large and complex datasets [22]. In general, as suggested by the literature, challenge resides in (a) high volumes of data making the aforementioned legacy algorithms reach unrealistic computation times, and (b) unm
Data Loading...
