Technical note: Some properties of splitting criteria

被引：192

作者：

Breiman, L

机构：

[1] Univ of California, Berkeley

来源：

MACHINE LEARNING | 1996年 / 24卷 / 01期

关键词：

trees; classification; splits;

D O I：

10.1023/A:1018094028462

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Various criteria have been proposed for deciding which split is best at a given node of a binary classification tree. Consider the question. given a goodness-of-split criterion and the class populations of the instances at a node, what distribution of the instances between the two children nodes maximizes the goodness-of-split criterion? The answers reveal an interesting distinction between the gini and entropy criterion.

引用

页码：41 / 47

页数：7

共 7 条

[1]

[Anonymous], 1993, P 13 INT JOINT C ART, DOI DOI 10.1109/TKDE.2011.181

[2]

Breiman L., 1984, Classification and Regression Trees, DOI DOI 10.2307/2530946

[3] A FURTHER COMPARISON OF SPLITTING RULES FOR DECISION-TREE INDUCTION [J].

BUNTINE, W ;

NIBLETT, T .

MACHINE LEARNING, 1992, 8 (01) :75-85

[4]

FAYYAD U, 1991, THESIS U MICHIGAN

[5]

FAYYAD UM, 1990, PROCEEDINGS : EIGHTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOLS 1 AND 2, P749

[6]

FAYYAD VM, 1992, P 10 NAT C AI AAAI 9, P104

[7] BLOCK DIAGRAMS AND SPLITTING CRITERIA FOR CLASSIFICATION TREES [J].

TAYLOR, PC ;

SILVERMAN, BW .

STATISTICS AND COMPUTING, 1993, 3 (04) :147-161

← 1 →