Model selection for dynamic processes

S Wu; PA Flach

Model selection for dynamic processes

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

Abstract

In machine learning, ROC (Receiver Operating Characteristic) analysis is widely used in model selection when we consider both class distribution and misclassification costs that must be given at test time. In this paper we consider the case of a dynamic process, such that the class distributions are different in different time periods or states. The main problem is then to decide when to change models according to the different states of the generating process. In this paper we use a control chart to choose models for the process when misclassification costs are considered. Four strategies are considered and model selection approaches are discussed.

Translated title of the contribution	Model selection for dynamic processes
Original language	English
Title of host publication	Unknown
Editors	M. Bohanec, B. Kasek, N. Lavrac, D. Mladenic
Publisher	University of Helsinki
Pages	168 - 173
Number of pages	5
Publication status	Published - Aug 2002

Bibliographical note

Conference Proceedings/Title of Journal: ECML/PKDD'02 workshop on Integration and Collaboration Aspects of Data Mining, Decision Support and Meta-Learning

Access to Document

http://www.cs.bris.ac.uk/Publications/pub_info.jsp?id=1000660

Handle.net

Persistent link

Cite this

@inproceedings{6c2b0783ced04eacac7b1a669b1abb57,

title = "Model selection for dynamic processes",

abstract = "In machine learning, ROC (Receiver Operating Characteristic) analysis is widely used in model selection when we consider both class distribution and misclassification costs that must be given at test time. In this paper we consider the case of a dynamic process, such that the class distributions are different in different time periods or states. The main problem is then to decide when to change models according to the different states of the generating process. In this paper we use a control chart to choose models for the process when misclassification costs are considered. Four strategies are considered and model selection approaches are discussed. ",

author = "S Wu and PA Flach",

note = "Conference Proceedings/Title of Journal: ECML/PKDD'02 workshop on Integration and Collaboration Aspects of Data Mining, Decision Support and Meta-Learning",

year = "2002",

month = aug,

language = "English",

pages = "168 -- 173",

editor = "M. Bohanec and B. Kasek and N. Lavrac and D. Mladenic",

booktitle = "Unknown",

publisher = "University of Helsinki",

address = "Finland",

}

TY - GEN

T1 - Model selection for dynamic processes

AU - Wu, S

AU - Flach, PA

N1 - Conference Proceedings/Title of Journal: ECML/PKDD'02 workshop on Integration and Collaboration Aspects of Data Mining, Decision Support and Meta-Learning

PY - 2002/8

Y1 - 2002/8

N2 - In machine learning, ROC (Receiver Operating Characteristic) analysis is widely used in model selection when we consider both class distribution and misclassification costs that must be given at test time. In this paper we consider the case of a dynamic process, such that the class distributions are different in different time periods or states. The main problem is then to decide when to change models according to the different states of the generating process. In this paper we use a control chart to choose models for the process when misclassification costs are considered. Four strategies are considered and model selection approaches are discussed.

AB - In machine learning, ROC (Receiver Operating Characteristic) analysis is widely used in model selection when we consider both class distribution and misclassification costs that must be given at test time. In this paper we consider the case of a dynamic process, such that the class distributions are different in different time periods or states. The main problem is then to decide when to change models according to the different states of the generating process. In this paper we use a control chart to choose models for the process when misclassification costs are considered. Four strategies are considered and model selection approaches are discussed.

M3 - Conference Contribution (Conference Proceeding)

SP - 168

EP - 173

BT - Unknown

A2 - Bohanec, M.

A2 - Kasek, B.

A2 - Lavrac, N.

A2 - Mladenic, D.

PB - University of Helsinki

ER -

Model selection for dynamic processes

Abstract

Bibliographical note

Access to Document

Handle.net

Fingerprint

Cite this