论文信息 - Bias in information-based measures in decision tree induction

Bias in information-based measures in decision tree induction

A fresh look is taken at the problem of bias in information-based attribute selection measures, used in the induction of decision trees. The approach uses statistical simulation techniques to demonstrate that the usual measures such as information gain, gain ratio, and a new measure recently proposed by Lopez de Mantaras (1991) are all biased in favour of attributes with large numbers of values. It is concluded that approaches which utilise the chi-square distribution are preferable because they compensate automatically for differences between attributes in the number of levels they take.

Wei Zhong Liu | A. White | W. Liu

[1] Solomon Kullback,et al. Information Theory and Statistics , 1960 .

[2] Elwyn Edwards. Information transmission : an introductory guide to the application of the theory of information to the human sciences , 1964 .

[3] G. Keppel,et al. Design and Analysis: A Researcher's Handbook , 1976 .

[4] W. W. Daniel. Applied Nonparametric Statistics , 1979 .

[5] Ivan Bratko,et al. Experiments in automatic learning of medical diagnostic rules , 1984 .

[6] John Mingers,et al. Expert Systems—Rule Induction with Statistical Data , 1987 .

[7] J. R. Quilan. Decision trees and multi-valued attributes , 1988 .

[8] John Mingers,et al. An Empirical Comparison of Selection Measures for Decision-Tree Induction , 1989, Machine Learning.

[9] Wei Zhong Liu,et al. The Importance of Attribute Selection Measures in Decision Tree Induction , 1994, Machine Learning.

[10] J. Ross Quinlan,et al. Induction of Decision Trees , 1986, Machine Learning.

[11] Ramón López de Mántaras,et al. A distance-based attribute selection measure for decision tree induction , 1991, Machine Learning.