Space decomposition in data mining - A clustering approach

Oded Maimon, Lior Rokach, Inbal Lavi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Decomposition may divide the database horizontally (subsets of rows or tuples) or vertically. It may be aimed at minimizing space and time needed for the classification of a dataset (e.g. sampling, windowing) or rather attempt to improve accuracy (e.g. bagging, boosting). This paper presents a horizontal space-decomposition algorithm, exploiting the K-means clustering algorithm. It is aimed at decreasing error rate compared to the simple classifier embedded in it while being rather understandable.
Original languageEnglish
Title of host publication22nd Convention of Electrical and Electronics Engineers in Israel, Proceedings
PublisherInstitute of Electrical and Electronics Engineers
Pages101-104
Number of pages4
ISBN (Electronic)0780376935
DOIs
StatePublished - 1 Jan 2002
Externally publishedYes
Event22nd Convention of Electrical and Electronics Engineers in Israel - Tel-Aviv, Israel
Duration: 1 Dec 2002 → …

Publication series

NameIEEE Convention of Electrical and Electronics Engineers in Israel, Proceedings
Volume2002-January

Conference

Conference22nd Convention of Electrical and Electronics Engineers in Israel
Country/TerritoryIsrael
CityTel-Aviv
Period1/12/02 → …

Keywords

  • Data mining
  • Induction generators
  • Probability distribution
  • Euclidean distance
  • very large databases
  • pattern clustering
  • database theory

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Space decomposition in data mining - A clustering approach'. Together they form a unique fingerprint.

Cite this