Subspace and projected clustering: Experimental evaluation and analysis

Gabriela Moise*, Arthur Zimek, Peer Kröger, Hans Peter Kriegel, Jörg Sander

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

Subspace and projected clustering have emerged as a possible solution to the challenges associated with clustering in high-dimensional data. Numerous subspace and projected clustering techniques have been proposed in the literature. A comprehensive evaluation of their advantages and disadvantages is urgently needed. In this paper, we evaluate systematically state-of-the-art subspace and projected clustering techniques under a wide range of experimental settings. We discuss the observed performance of the compared techniques, and we make recommendations regarding what type of techniques are suitable for what kind of problems.

Original languageEnglish
JournalKnowledge and Information Systems
Volume21
Issue number3
Pages (from-to)299-326
ISSN0219-1377
DOIs
Publication statusPublished - Dec 2009
Externally publishedYes

Keywords

  • Projected clustering
  • Subspace clustering

Cite this

Moise, Gabriela ; Zimek, Arthur ; Kröger, Peer ; Kriegel, Hans Peter ; Sander, Jörg. / Subspace and projected clustering : Experimental evaluation and analysis. In: Knowledge and Information Systems. 2009 ; Vol. 21, No. 3. pp. 299-326.
@article{369895eb282e4dad9d24deb773e33630,
title = "Subspace and projected clustering: Experimental evaluation and analysis",
abstract = "Subspace and projected clustering have emerged as a possible solution to the challenges associated with clustering in high-dimensional data. Numerous subspace and projected clustering techniques have been proposed in the literature. A comprehensive evaluation of their advantages and disadvantages is urgently needed. In this paper, we evaluate systematically state-of-the-art subspace and projected clustering techniques under a wide range of experimental settings. We discuss the observed performance of the compared techniques, and we make recommendations regarding what type of techniques are suitable for what kind of problems.",
keywords = "Projected clustering, Subspace clustering",
author = "Gabriela Moise and Arthur Zimek and Peer Kr{\"o}ger and Kriegel, {Hans Peter} and J{\"o}rg Sander",
year = "2009",
month = "12",
doi = "10.1007/s10115-009-0226-y",
language = "English",
volume = "21",
pages = "299--326",
journal = "Knowledge and Information Systems",
issn = "0219-1377",
publisher = "Springer",
number = "3",

}

Subspace and projected clustering : Experimental evaluation and analysis. / Moise, Gabriela; Zimek, Arthur; Kröger, Peer; Kriegel, Hans Peter; Sander, Jörg.

In: Knowledge and Information Systems, Vol. 21, No. 3, 12.2009, p. 299-326.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Subspace and projected clustering

T2 - Experimental evaluation and analysis

AU - Moise, Gabriela

AU - Zimek, Arthur

AU - Kröger, Peer

AU - Kriegel, Hans Peter

AU - Sander, Jörg

PY - 2009/12

Y1 - 2009/12

N2 - Subspace and projected clustering have emerged as a possible solution to the challenges associated with clustering in high-dimensional data. Numerous subspace and projected clustering techniques have been proposed in the literature. A comprehensive evaluation of their advantages and disadvantages is urgently needed. In this paper, we evaluate systematically state-of-the-art subspace and projected clustering techniques under a wide range of experimental settings. We discuss the observed performance of the compared techniques, and we make recommendations regarding what type of techniques are suitable for what kind of problems.

AB - Subspace and projected clustering have emerged as a possible solution to the challenges associated with clustering in high-dimensional data. Numerous subspace and projected clustering techniques have been proposed in the literature. A comprehensive evaluation of their advantages and disadvantages is urgently needed. In this paper, we evaluate systematically state-of-the-art subspace and projected clustering techniques under a wide range of experimental settings. We discuss the observed performance of the compared techniques, and we make recommendations regarding what type of techniques are suitable for what kind of problems.

KW - Projected clustering

KW - Subspace clustering

U2 - 10.1007/s10115-009-0226-y

DO - 10.1007/s10115-009-0226-y

M3 - Journal article

AN - SCOPUS:71949123741

VL - 21

SP - 299

EP - 326

JO - Knowledge and Information Systems

JF - Knowledge and Information Systems

SN - 0219-1377

IS - 3

ER -