摘要
针对ID3算法用信息增益作为在各级非叶节点上选择属性的标准的局限性,结合统计学独立检验思想,给出一种新的属性依赖性和重要性定义,以新的属性重要性为启发式信息设计决策树规则提取算法。实例分析的结果表明,该算法能提取更为简洁有效的决策规则。
A new attribute dependency and significance were defined with the independent eheck theory of the statistical aiming at the disadvantages of the standard for choosing the attributes of the branch nodes with the information gain in the ID3 algorithm. An algorithm for rules extraction of decision tree was designed. The new attribute significance was used as the heuristic information in which. The experiment and comparison show that the algorithm provides more precise and simple decision tree.
出处
《河北工程大学学报（自然科学版）》
CAS
2008年第1期108-110,共3页
Journal of Hebei University of Engineering:Natural Science Edition
关键词
粗糙集
属性依赖性
决策树
规则提取
rough set
attribute dependency
decision tree
rules extraction
