Handling Time Changing Data with Adaptive Very Fast Decision
Rules

D 2012

Handling Time Changing Data with Adaptive Very Fast Decision Rules

KOSINA, Petr and Joao GAMA

Basic information

Original name

Handling Time Changing Data with Adaptive Very Fast Decision Rules

Authors

KOSINA, Petr (203 Czech Republic, guarantor, belonging to the institution) and Joao GAMA (620 Portugal)

Edition

Berlin / Heidelberg, Machine Learning and Knowledge Discovery in Databases ECML/PKDD, p. 827-842, 16 pp. 2012

Publisher

Springer Berlin / Heidelberg

Other information

Language

English

Type of outcome

Proceedings paper

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

Czech Republic

Confidentiality degree

is not subject to a state or trade secret

Publication form

printed version "print"

References:

URL

Impact factor

Impact factor: 0.402 in 2005

RIV identification code

RIV/00216224:14330/12:00061019

Organization unit

Faculty of Informatics

ISBN

978-3-642-33459-7

ISSN

DOI

https://doi.org/10.1007/978-3-642-33460-3_58

Keywords in English

Data Streams; Decision Rules; Concept Drift

Abstract

In the original language

Data streams are usually characterized by changes in the underlying distribution generating data. Therefore algorithms designed to work with data streams should be able to detect changes and quickly adapt the decision model. Rules are one of the most interpretable and flexible models for data mining prediction tasks. In this paper we present the Adaptive Very Fast Decision Rules (AVFDR), an on-line, any-time and one-pass algorithm for learning decision rules in the context of time changing data. AVFDR can learn ordered and unordered rule sets. It is able to adapt the decision model via incremental induction and specialization of rules. Detecting local drifts takes advantage of the modularity of rule sets. In AVFDR, each individual rule monitors the evolution of performance metrics to detect concept drift. AVFDR prunes rules that detect drift. This explicit change detection mechanism provides useful information about the dynamics of the process generating data, faster adaption to changes and generates compact rule sets. The experimental evaluation shows this method is able to learn fast and compact rule sets from evolving streams in comparison to alternative methods.

Links

LA09016, research and development project

Name: Účast ČR v European Research Consortium for Informatics and Mathematics (ERCIM) (Acronym: ERCIM)

Investor: Ministry of Education, Youth and Sports of the CR, Czech Republic membership in the European Research Consortium for Informatics and Mathematics

MUNI/A/0758/2011, interní kód MU

Name: Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity (Acronym: SKOMU)

Investor: Masaryk University, Category A

Cite

KOSINA, Petr and Joao GAMA. Handling Time Changing Data with Adaptive Very Fast Decision Rules. In Peter Flach, Tijl De Bie, Nello Cristianini. Machine Learning and Knowledge Discovery in Databases ECML/PKDD. Berlin / Heidelberg: Springer Berlin / Heidelberg, 2012, p. 827-842. ISBN 978-3-642-33459-7. Available from: https://doi.org/10.1007/978-3-642-33460-3_58.

@inproceedings{993086,
   author = {Kosina, Petr and Gama, Joao},
   address = {Berlin / Heidelberg},
   booktitle = {Machine Learning and Knowledge Discovery in Databases ECML/PKDD},
   doi = {https://doi.org/10.1007/978-3-642-33460-3_58},
   editor = {Peter Flach, Tijl De Bie, Nello Cristianini},
   keywords = {Data Streams; Decision Rules; Concept Drift},
   howpublished = {tištěná verze "print"},
   language = {eng},
   location = {Berlin / Heidelberg},
   isbn = {978-3-642-33459-7},
   pages = {827-842},
   publisher = {Springer Berlin / Heidelberg},
   title = {Handling Time Changing Data with Adaptive Very Fast Decision Rules},
   url = {http://dx.doi.org/10.1007/978-3-642-33460-3_58},
   year = {2012}
}

TY  - CONF
ID  - 993086
AU  - Kosina, Petr - Gama, Joao
PY  - 2012
TI  - Handling Time Changing Data with Adaptive Very Fast Decision Rules
PB  - Springer Berlin / Heidelberg
CY  - Berlin / Heidelberg
SN  - 9783642334597
KW  - Data Streams
KW  - Decision Rules
KW  - Concept Drift
UR  - http://dx.doi.org/10.1007/978-3-642-33460-3_58
L2  - http://dx.doi.org/10.1007/978-3-642-33460-3_58
N2  - Data streams are usually characterized by changes in the underlying distribution generating data. Therefore algorithms designed to work with data streams should be able to detect changes and quickly adapt the decision model. Rules are one of the most interpretable and flexible models for data mining prediction tasks. In this paper we present the Adaptive Very Fast Decision Rules (AVFDR), an on-line, any-time and one-pass algorithm for learning decision rules in the context of time changing data. AVFDR can learn ordered and unordered rule sets. It is able to adapt the decision model via incremental induction and specialization of rules. Detecting local drifts takes advantage of the modularity of rule sets. In AVFDR, each individual rule monitors the evolution of performance metrics to detect concept drift. AVFDR prunes rules that detect drift. This explicit change detection mechanism provides useful information about the dynamics of the process generating data, faster adaption to changes and generates compact rule sets. The experimental evaluation shows this method is able to learn fast and compact rule sets from evolving streams in comparison to alternative methods.
ER  -

KOSINA, Petr and Joao GAMA. Handling Time Changing Data with Adaptive Very Fast Decision Rules. In Peter Flach, Tijl De Bie, Nello Cristianini. \textit{Machine Learning and Knowledge Discovery in Databases ECML/PKDD}. Berlin / Heidelberg: Springer Berlin / Heidelberg, 2012, p.~827-842. ISBN~978-3-642-33459-7. Available from: https://doi.org/10.1007/978-3-642-33460-3\_{}58.

Přehled o publikaci