An Improved Algorithm Based on CART Decision
Hao Zhong-xiao · Harbin Ligong Daxue xuebao · 2009
In view of basis on analyzing insufficient of the classifacation and regression trees(CART) algorithm,which requests the choosen attribute is continual and having order,and onlycan produce two children nodes,based on CART algorithm theory,this paper uses A-CART the algorithm,makes up the CART method insufficiency,namely the method which selects the discrete and disorder attribute to accomplish split point of decision tree,after splitting it may have many split node.It has designed a split function for the algorithm,and has given the concrete application method through an example.