摘要
文章从决策树自身结构出发,总结出了三条后剪枝规则,并通过决策树的树形结构和规则集两种等价的表示方式,讨论了剪枝技术的实现。该方法无需任何其它先验知识,简洁高效,易于实现,易于实现图形化可视界面,且100%拟合训练数据集,可以独立使用,也可以作为其它剪枝算法的基础,从而加快其它算法的收敛速度。
This paper summarized up three post-pruning rules based on decision tree＇s structural characteristics,and discussed the implementation algorithm using the decision tree＇s structure and its equivalent rule set.The technique proposed in this paper,does not need any other prior knowledge.It is simple and efficient,and easy to implement,easy to develop a graphical visual interface.Moreover it fits the training data set perfectly,can be used as an independent post-pruning technique as well as a basis for other algorithms,thus speeding up their convergence.
出处
《电脑与信息技术》
2010年第4期1-4,16,共5页
Computer and Information Technology
关键词
决策树
分类规则
后剪枝技术
decision tree
classification rules
post-pruning technique
