Behave or Be Detected! Identifying Outlier Sequences by Their Group Cohesion

Tatusch, Martha; Klassen, Gerhard; Conrad, Stefan

doi:10.1007/978-3-030-59065-9_26

Cited by 2 publications

(4 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our evaluation measures can not only be used for the overtime stability evaluation of clusterings, but also for further analyses such as parameter selection or outlier detection [50,52,53]. Therefore, for example the part of CLOSE, where subsequences are evaluated, can be used.…”

Section: Applicationsmentioning

confidence: 99%

“…In order to counteract this circumstance, the jaccard index Fig. 4 Example for cluster transitions of time series T a , .., T e over time [52] can be used, as proposed in [52]. Therefore, the temporal cluster union of two clusters C t i ,a , C t j ,b has to be introduced first:…”

Section: Applicationsmentioning

confidence: 99%

“…with k ∈ [1, j − i] again being the number of timestamps between t i and t j where the considered time series exists [52]. There is no need to normalize the score to an interval of [0, 1] by averaging it, as the sum of all weightings of a subsequence's timestamps is always 1 due the division by the Gauss's Formula.…”

Section: Applicationsmentioning

confidence: 99%

“…Further, we are presenting an outlier detection algorithm [50] which is an application of CLOSE. We give two variants [52] of the procedure which focus on cluster transitions and therefore are capable to detect a new sort of outliers, which are based on the behavior of time series in relation to its cluster peers. The implementation of the approaches as well as the generated data sets are available on Github 2 .…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Cluster-based stability evaluation in time series data sets

2022

Self Cite

View full text Add to dashboard Cite

In modern data analysis, time is often considered just another feature. Yet time has a special role that is regularly overlooked. Procedures are usually only designed for time-independent data and are therefore often unsuitable for the temporal aspect of the data. This is especially the case for clustering algorithms. Although there are a few evolutionary approaches for time-dependent data, the evaluation of these and therefore the selection is difficult for the user. In this paper, we present a general evaluation measure that examines clusterings with respect to their temporal stability and thus provides information about the achieved quality. For this purpose, we examine the temporal stability of time series with respect to their cluster neighbors, the temporal stability of clusters with respect to their composition, and finally conclude on the temporal stability of the entire clustering. We summarise these components in a parameter-free toolkit that we call Cluster Over-Time Stability Evaluation (CLOSE). In addition to that we present a fuzzy variant which we call FCSETS (Fuzzy Clustering Stability Evaluation of Time Series). These toolkits enable a number of advanced applications. One of these is parameter selection for any type of clustering algorithm. We demonstrate parameter selection as an example and evaluate results of classical clustering algorithms against a well-known evolutionary clustering algorithm. We then introduce a method for outlier detection in time series data based on CLOSE. We demonstrate the practicality of our approaches on three real world data sets and one generated data set.

show abstract

Section: Applicationsmentioning

confidence: 99%

Section: Applicationsmentioning

confidence: 99%

Section: Applicationsmentioning

confidence: 99%