Abstract:
This paper develops a decision tree classifier SSC(similarity of same and consistent) for multi-valued and multi-labeled data, improves on MMC’s formula for measuring the similarity of label-sets to determine the goodness of splitting attributes. It proposes a new measure approach considering both same and consistent features of label-sets. The experiment shows SSC has improved accuracy of MMC.
Key words:
classification,
decision tree,
multi-valued attribute,
multi-labeled data,
similarity
摘要: 提出了一种多值属性和多类标数据的决策树算法(SSC),在MMC算法中,对用孩子结点的类标集相似度来评定结点属性分类效果的计算方法进行了改进,综合考虑集合的同一性和一致性,提出了相似度评定方法,使类标集相似度的计算更加全面和准确。实验证明该算法的分类效果优于MMC算法。
关键词:
分类,
决策树,
多值属性,
多类标数据,
相似度
CLC Number:
ZHAO Rui; LI Hong. Algorithm of Multi-valued Attribute and Multi-labeled Data Decision Tree[J]. Computer Engineering, 2007, 33(13): 87-89.
赵 蕊;李 宏. 一种多值属性和多类标数据的决策树算法[J]. 计算机工程, 2007, 33(13): 87-89.