Visualizing Bagged Decision Trees

J. Sunil Rao; William J.E. Potts

Visualizing Bagged Decision Trees

J. Sunil Rao, William J.E. Potts

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We present a visual tablet for exploring the nature of a bagged decision tree (Breiman [1996]). Aggregating classifiers over bootstrap datasets (bagging) can result in greatly improved prediction accuracy. Bagging is motivated as a variance reduction technique, but it is considered a black box with respect to interpretation. Current research seeking to explain why bagging works has focused on different bias/variance decompositions of prediction error. We show that bagging's complexity can be better understood by a simple graphical technique that allows visualizing the bagged decision boundary in low-dimensional situations. We then show that bagging can be heuristically motivated as a method to enhance local adaptivity of the boundary. Some simulated examples are presented to illustrate the technique.

Original language	English (US)
Title of host publication	Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997
Editors	David Heckerman, Heikki Mannila, Daryl Pregibon, Ramasamy Uthurusamy
Publisher	AAAI press
Pages	243-246
Number of pages	4
ISBN (Electronic)	1577350278, 9781577350279
State	Published - 1997
Externally published	Yes
Event	3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997 - Newport Beach, United States Duration: Aug 14 1997 → Aug 17 1997

Publication series

Name	Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997

Conference

Conference	3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997
Country/Territory	United States
City	Newport Beach
Period	8/14/97 → 8/17/97

Bibliographical note

Publisher Copyright:
Copyright © 1997, American Association for Artificial Intelligence (www.aaai.org). All rights reserved.

Find It

Find It at U of M Libraries

Cite this

Visualizing Bagged Decision Trees. / Rao, J. Sunil; Potts, William J.E.
Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997. ed. / David Heckerman; Heikki Mannila; Daryl Pregibon; Ramasamy Uthurusamy. AAAI press, 1997. p. 243-246 (Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Rao, JS & Potts, WJE 1997, Visualizing Bagged Decision Trees. in D Heckerman, H Mannila, D Pregibon & R Uthurusamy (eds), Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997. Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997, AAAI press, pp. 243-246, 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997, Newport Beach, United States, 8/14/97.

Rao, J. Sunil ; Potts, William J.E. / Visualizing Bagged Decision Trees. Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997. editor / David Heckerman ; Heikki Mannila ; Daryl Pregibon ; Ramasamy Uthurusamy. AAAI press, 1997. pp. 243-246 (Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997).

@inproceedings{33454cd9cc9746cbb04fd09058426aaf,

title = "Visualizing Bagged Decision Trees",

abstract = "We present a visual tablet for exploring the nature of a bagged decision tree (Breiman [1996]). Aggregating classifiers over bootstrap datasets (bagging) can result in greatly improved prediction accuracy. Bagging is motivated as a variance reduction technique, but it is considered a black box with respect to interpretation. Current research seeking to explain why bagging works has focused on different bias/variance decompositions of prediction error. We show that bagging's complexity can be better understood by a simple graphical technique that allows visualizing the bagged decision boundary in low-dimensional situations. We then show that bagging can be heuristically motivated as a method to enhance local adaptivity of the boundary. Some simulated examples are presented to illustrate the technique.",

author = "Rao, {J. Sunil} and Potts, {William J.E.}",

note = "Publisher Copyright: Copyright {\textcopyright} 1997, American Association for Artificial Intelligence (www.aaai.org). All rights reserved.; 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997 ; Conference date: 14-08-1997 Through 17-08-1997",

year = "1997",

language = "English (US)",

series = "Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997",

publisher = "AAAI press",

pages = "243--246",

editor = "David Heckerman and Heikki Mannila and Daryl Pregibon and Ramasamy Uthurusamy",

booktitle = "Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997",

}

TY - GEN

T1 - Visualizing Bagged Decision Trees

AU - Rao, J. Sunil

AU - Potts, William J.E.

PY - 1997

Y1 - 1997

N2 - We present a visual tablet for exploring the nature of a bagged decision tree (Breiman [1996]). Aggregating classifiers over bootstrap datasets (bagging) can result in greatly improved prediction accuracy. Bagging is motivated as a variance reduction technique, but it is considered a black box with respect to interpretation. Current research seeking to explain why bagging works has focused on different bias/variance decompositions of prediction error. We show that bagging's complexity can be better understood by a simple graphical technique that allows visualizing the bagged decision boundary in low-dimensional situations. We then show that bagging can be heuristically motivated as a method to enhance local adaptivity of the boundary. Some simulated examples are presented to illustrate the technique.

AB - We present a visual tablet for exploring the nature of a bagged decision tree (Breiman [1996]). Aggregating classifiers over bootstrap datasets (bagging) can result in greatly improved prediction accuracy. Bagging is motivated as a variance reduction technique, but it is considered a black box with respect to interpretation. Current research seeking to explain why bagging works has focused on different bias/variance decompositions of prediction error. We show that bagging's complexity can be better understood by a simple graphical technique that allows visualizing the bagged decision boundary in low-dimensional situations. We then show that bagging can be heuristically motivated as a method to enhance local adaptivity of the boundary. Some simulated examples are presented to illustrate the technique.

UR - http://www.scopus.com/inward/record.url?scp=85172416531&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85172416531&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85172416531

T3 - Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997

SP - 243

EP - 246

BT - Proceedings - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997

A2 - Heckerman, David

A2 - Mannila, Heikki

A2 - Pregibon, Daryl

A2 - Uthurusamy, Ramasamy

PB - AAAI press

T2 - 3rd International Conference on Knowledge Discovery and Data Mining, KDD 1997

Y2 - 14 August 1997 through 17 August 1997

ER -

Visualizing Bagged Decision Trees

Abstract

Publication series

Conference

Bibliographical note

Find It

Other files and links

Fingerprint

Cite this