Instance space analysis for unsupervised outlier detection

Sevvandi Kandanaarachchi; Mario A. Muñoz; Kate Smith-Miles

Instance space analysis for unsupervised outlier detection

Sevvandi Kandanaarachchi, Mario A. Muñoz, Kate Smith-Miles

Econometrics & Business Statistics

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Other › peer-review

Abstract

The success of various methods for unsupervised outlier detection t depends on how well their definition of an outlier matches the dataset properties. Given that each definition, and hence each t method, has strengths and weaknesses, measuring those properties could help us to predict the most suitable method for the dataset at hand. In this paper, we construct and validate a set of meta-features that measure such properties. We then conduct the first instance space analysis for unsupervised outlier detection methods based on the meta-features. The analysis provides insights into the methods’ strengths and weaknesses, and facilitates the recommendation of an appropriate method with good accuracy. (

Original language	English
Title of host publication	Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019)
Editors	Eirini Ntoutsi, Erich Schubert, Arthur Zimek, Albrecht Zimmermann
Place of Publication	Germany
Publisher	Rheinisch-Westfaelische Technische Hochschule Aachen
Number of pages	9
Publication status	Published - 2019
Event	Evaluation and Experimental Design in Data Mining and Machine Learning 2019 - Hyatt Regency Calgary, Calgary, Canada Duration: 4 May 2019 → … Conference number: 1st https://imada.sdu.dk/Research/EDML/ (Workshop website) http://ceur-ws.org/Vol-2436/ (Proceedings)

Publication series

Name	CEUR Workshop Proceedings
Volume	2436
ISSN (Print)	1613-0073

Workshop

Workshop	Evaluation and Experimental Design in Data Mining and Machine Learning 2019
Abbreviated title	EDML 2019
Country/Territory	Canada
City	Calgary
Period	4/05/19 → …
Other	Workshop at the SIAM International Conference on Data Mining (SDM19)
Internet address	https://imada.sdu.dk/Research/EDML/ (Workshop website) http://ceur-ws.org/Vol-2436/ (Proceedings)

Access to Document

285878705-oaFinal published version, 2.2 MB

http://ceur-ws.org/Vol-2436/article_4.pdf

Cite this

Kandanaarachchi, S., Muñoz, M. A., & Smith-Miles, K. (2019). Instance space analysis for unsupervised outlier detection. In E. Ntoutsi, E. Schubert, A. Zimek, & A. Zimmermann (Eds.), Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019) (CEUR Workshop Proceedings; Vol. 2436). Rheinisch-Westfaelische Technische Hochschule Aachen. http://ceur-ws.org/Vol-2436/article_4.pdf

Kandanaarachchi, Sevvandi ; Muñoz, Mario A. ; Smith-Miles, Kate. / Instance space analysis for unsupervised outlier detection. Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019). editor / Eirini Ntoutsi ; Erich Schubert ; Arthur Zimek ; Albrecht Zimmermann. Germany : Rheinisch-Westfaelische Technische Hochschule Aachen, 2019. (CEUR Workshop Proceedings).

@inproceedings{4ac55b4ca313404a8a0c524be18a9bc0,

title = "Instance space analysis for unsupervised outlier detection",

abstract = "The success of various methods for unsupervised outlier detection t depends on how well their definition of an outlier matches the dataset properties. Given that each definition, and hence each t method, has strengths and weaknesses, measuring those properties could help us to predict the most suitable method for the dataset at hand. In this paper, we construct and validate a set of meta-features that measure such properties. We then conduct the first instance space analysis for unsupervised outlier detection methods based on the meta-features. The analysis provides insights into the methods{\textquoteright} strengths and weaknesses, and facilitates the recommendation of an appropriate method with good accuracy. (",

author = "Sevvandi Kandanaarachchi and Mu{\~n}oz, {Mario A.} and Kate Smith-Miles",

year = "2019",

language = "English",

series = "CEUR Workshop Proceedings",

publisher = "Rheinisch-Westfaelische Technische Hochschule Aachen",

editor = "Eirini Ntoutsi and Erich Schubert and Arthur Zimek and Albrecht Zimmermann",

booktitle = "Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019)",

note = "Evaluation and Experimental Design in Data Mining and Machine Learning 2019, EDML 2019 ; Conference date: 04-05-2019",

url = "https://imada.sdu.dk/Research/EDML/, http://ceur-ws.org/Vol-2436/",

}

Kandanaarachchi, S, Muñoz, MA & Smith-Miles, K 2019, Instance space analysis for unsupervised outlier detection. in E Ntoutsi, E Schubert, A Zimek & A Zimmermann (eds), Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019). CEUR Workshop Proceedings, vol. 2436, Rheinisch-Westfaelische Technische Hochschule Aachen, Germany, Evaluation and Experimental Design in Data Mining and Machine Learning 2019, Calgary, Alberta, Canada, 4/05/19. <http://ceur-ws.org/Vol-2436/article_4.pdf>

Instance space analysis for unsupervised outlier detection. / Kandanaarachchi, Sevvandi; Muñoz, Mario A.; Smith-Miles, Kate.
Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019). ed. / Eirini Ntoutsi; Erich Schubert; Arthur Zimek; Albrecht Zimmermann. Germany: Rheinisch-Westfaelische Technische Hochschule Aachen, 2019. (CEUR Workshop Proceedings; Vol. 2436).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Other › peer-review

TY - GEN

T1 - Instance space analysis for unsupervised outlier detection

AU - Kandanaarachchi, Sevvandi

AU - Muñoz, Mario A.

AU - Smith-Miles, Kate

N1 - Conference code: 1st

PY - 2019

Y1 - 2019

N2 - The success of various methods for unsupervised outlier detection t depends on how well their definition of an outlier matches the dataset properties. Given that each definition, and hence each t method, has strengths and weaknesses, measuring those properties could help us to predict the most suitable method for the dataset at hand. In this paper, we construct and validate a set of meta-features that measure such properties. We then conduct the first instance space analysis for unsupervised outlier detection methods based on the meta-features. The analysis provides insights into the methods’ strengths and weaknesses, and facilitates the recommendation of an appropriate method with good accuracy. (

AB - The success of various methods for unsupervised outlier detection t depends on how well their definition of an outlier matches the dataset properties. Given that each definition, and hence each t method, has strengths and weaknesses, measuring those properties could help us to predict the most suitable method for the dataset at hand. In this paper, we construct and validate a set of meta-features that measure such properties. We then conduct the first instance space analysis for unsupervised outlier detection methods based on the meta-features. The analysis provides insights into the methods’ strengths and weaknesses, and facilitates the recommendation of an appropriate method with good accuracy. (

UR - http://www.scopus.com/inward/record.url?scp=85072753127&partnerID=8YFLogxK

M3 - Conference Paper

AN - SCOPUS:85072753127

T3 - CEUR Workshop Proceedings

BT - Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019)

A2 - Ntoutsi, Eirini

A2 - Schubert, Erich

A2 - Zimek, Arthur

A2 - Zimmermann, Albrecht

PB - Rheinisch-Westfaelische Technische Hochschule Aachen

CY - Germany

T2 - Evaluation and Experimental Design in Data Mining and Machine Learning 2019

Y2 - 4 May 2019

ER -

Kandanaarachchi S, Muñoz MA, Smith-Miles K. Instance space analysis for unsupervised outlier detection. In Ntoutsi E, Schubert E, Zimek A, Zimmermann A, editors, Proceedings of the 1st Workshop on Evaluation and Experimental Design in Data Mining and Machine Learning co-located with SIAM International Conference on Data Mining (SDM 2019). Germany: Rheinisch-Westfaelische Technische Hochschule Aachen. 2019. (CEUR Workshop Proceedings).

Instance space analysis for unsupervised outlier detection

Abstract

Publication series

Workshop

Access to Document

Other files and links

Cite this