New construction approach of decision tree
Zhangyan Xu · Computer Engineering and Applications Journal · 2009
Decision tree is an important tool for data mining.However,the design of optimal decision tree is proved to be a NPcomplete problem.A construction approach of decision tree based on association rule is proposed.Firstly,approximate exact rule with high confidence is defined.Then,an algorithm for mining approximate exact rules is proposed.Thirdly,approximate exact rules are used to generate new attributes.Fourthly,the method on new attribute evaluation is discussed.Finally,together with original attributes,new generated attributes are used for construction of decision tree.Experimental results show the accuracy of the proposed method is high.