Fuzzy data mining for discovering changes in association rules over time

Wai Ho Au, Chun Chung Chan

Research output: Journal article publicationConference articleAcademic researchpeer-review

14 Citations (Scopus)

Abstract

Association rule mining is an important topic in data mining research. Many algorithms have been developed for such task and they typically assume that the underlying associations hidden in the data are stable over time. However, in real world domains, it is possible that the data characteristics and hence the associations change significantly over time. Existing data mining algorithms have not taken the changes in associations into consideration and this can result in severe degradation of performance, especially when the discovered association rules are used for classification (prediction). Although the mining of changes in associations is an important problem because it is common that we need to predict the future based on the historical data in the past, existing data mining algorithms are not developed for this task. In this paper, we introduce a new fuzzy data mining technique to discover changes in association rules over time. Our approach mines fuzzy rules to represent the changes in association rules. Based on the discovered fuzzy rules, our approach is able to predict how the association rules will change in the future. The experimental results on a real-life database have shown that our approach is very effective in mining and predicting changes in association rules over time.
Original languageEnglish
Pages (from-to)890-895
Number of pages6
JournalIEEE International Conference on Fuzzy Systems
Volume2
Publication statusPublished - 31 Dec 2002
Event2002 IEEE International Conference on Fuzzy Systems: FUZZ-IEEE'02 - Honolulu, HI, United States
Duration: 12 May 200217 May 2002

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Software
  • Artificial Intelligence
  • Applied Mathematics

Cite this