Well-behaved evaluation functions for numerical attributes

Tapio Elomaa, Juho Rousu

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

Abstract

The class of well-behaved evaluation functions simplifies and makes efficient the handling of numerical attributes; for them it suffices to concentrate on the boundary points in searching for the optimal partition. This holds always for binary partitions and also for multisplits if only the function is cumulative in addition to being well-behaved. A large portion of the most important attribute evaluation functions are well-behaved. This paper surveys the class of well-behaved functions. As a case study, we examine the properties of C4.5's attribute evaluation functions. Our empirical experiments show that a very simple cumulative rectification to the poor bias of information gain significantly outperforms gain ratio.
Original languageEnglish
Title of host publicationFoundations of Intelligent Systems
Subtitle of host publication10th International Symposium, ISMIS '97
EditorsZbigniew W. Raś, Andrzej Skowron
Place of PublicationHeidelberg
PublisherSpringer
Pages147-156
ISBN (Electronic)978-3-540-69612-4
ISBN (Print)978-3-540-63614-4
DOIs
Publication statusPublished - 1997
MoE publication typeA4 Article in a conference publication
Event10th International Symposium, ISMIS '97 - Charlotte, United States
Duration: 15 Oct 199718 Oct 1997

Publication series

SeriesLecture Notes in Computer Science
Volume1325
ISSN0302-9743

Conference

Conference10th International Symposium, ISMIS '97
Country/TerritoryUnited States
CityCharlotte
Period15/10/9718/10/97

Fingerprint

Dive into the research topics of 'Well-behaved evaluation functions for numerical attributes'. Together they form a unique fingerprint.

Cite this