摘要
阐述了基于gini系数的决策树构造算法——CART算法。为了使算法能处理递增的数据,引入递增式学习方法,提出了一种改进算法。递增式学习的主要思想是测试函数提升。首先使用己有的数据用CART算法生成一棵决策树,然后使用递增的数据和递增式学习的方法来修改己有的决策树。最后从理论和实践两方面证明了改进算法的正确性和有效性。
A decision-tree arithmetic- CART based on gini is expatiated. To deal with the degrees data, a improved arithmetic combined with degrees study is advanced, it＇s main idea is function upgrade. First existing data is used to build a decision-tree by CART, then degrees data and degrees study are used to modify existing decision-tree. Finally proved correctness and validity of the improved arithmetic from theory and practice.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第7期1520-1522,共3页
Computer Engineering and Design
