摘要
提出了一种基于模糊化决策树的自适应分类算法 .介绍基于决策树的分类算法 ,指出训练样本分布不均匀或树剪枝操作都可能引起分类规则的不完全 ,导致分类出现“盲区” .引入决策树的模糊化方法及分支 (规则 )激活度的概念 ,给出一种新的自适应分类算法 .并用实例分析表明 ,该算法不仅解决了分类规则不完全的问题 ,而且也提高了决策树分类的精度及分类结果的可解释性 .
An adaptive classification algorithm based on fuzzy decision tree is proposed. First, decision tree classification is introduced. It is pointed out that classification rules are usually incomplete due to the distribution of samples and the tree pruning, which can produce “blind region” during classification. Then the fuzzification of decision tree is introduced and a new adaptive classification algorithm is proposed. Experimental results show that the new adaptive classification algorithm can solve the problem of “blind region” perfectly, and the precision and interpretability of the classification results are improved.
出处
《北京师范大学学报(自然科学版)》
CAS
CSCD
北大核心
2004年第5期582-587,共6页
Journal of Beijing Normal University(Natural Science)
基金
国家自然科学基金资助项目 (6 0 174 0 13)
教育部博士点基金资助项目 (2 0 0 2 0 0 2 70 13)
教育部科学技术重点项目(0 3184 )
国家"九七三"重大基础研究规划基金资助项目 (2 0 0 2CB312 2 0 0 )
关键词
分类算法
决策树
决策树的模糊化
分类规则的模糊化
近似推理
classification algorithm
decision tree
fuzzification of decision tree
fuzzification of classification rules
approximate reasoning