Bagging, Boosting, and C4.5

Breiman's bagging and Freund and Schapire's boosting are recent methods for improving the predictive power of classi er learning systems. Both form a set of classi ers that are combined by voting, bagging by generating replicated bootstrap samples of the data, and boosting by adjusting the weights of training instances. This paper reports results of applying both techniques to a system that learns decision trees and testing on a representative collection of datasets. While both approaches substantially improve predictive accuracy, boosting shows the greater bene t. On the other hand, boosting also produces severe degradation on some datasets. A small change to the way that boosting combines the votes of learned classi ers reduces this downside and also leads to slightly better results on most of the datasets considered.

• Type: Conference
• Year: 1996
• Where: AAAI
• Authors: J. Ross Quinlan
• Keywords: AAAI 1996, Boosting, Classi Ers, Intelligent Agents, Schapire's Boosting