Research And Application On Theory Of Decision Tree Based On Dynamic Fuzzy Lattice

With the studying of decision tree, the decision tree learning that only can describe the accurate characters is not adapted to the requirement describing the imprecise knowledge of a system. The world human being belong to and the domains issues within are dynamic. So in the real world, we hope the decision tree is capable of describing the dynamic fuzzy issues.We do elementary research for answering the question and get some production as follow:Firstly, we analysis decision tree and gain its dynamic and fuzzy characters. Using dynamic fuzzy lattice's describing system denotes decision tree. We accomplish the operation importing dynamic fuzzy lattice for decision tree. We bring forward the newly attribute chosen algorithm based on classified certain degree of condition attributes for decision attribute. According to it, we build dynamic fuzzy decision tree (DFDT). In DFDT, the parent nodes and child nodes plot out the instance set, each layer of DFDT does same also. We construe the relation of partition of instances set through establishing the dynamic fuzzy partition lattice. Secondly, We introduce the techniques dealing with the specific attributes in dynamic fuzzy decision tree. In this section, we also give the algorithm of multi-numeric attributes discretization. Then, with the summing-up of pruning methods of decision tree, we make certain the pruning strategy that adapts to dynamic fuzzy decision tree. At last, we research the application of dynamic fuzzy decision tree.All in all, features of the paper represent as follow:(1)Theory of dynamic fuzzy partition lattice riches the content of dynamic fuzzy set.(2)The reasonable describing about dynamic fuzzy decision tree from attributes treatment to building the tree and then pruning tree, and it provides a certain extent stated theory foundation for ulteriorly researching dynamic fuzzy decision tree and establishes a main concept frame of dynamic fuzzy decision tree.Although we got some summings-up, the work what we do is basic research. There are some questions that need be deeply researched. For example, the material application of dynamic fuzzy decision tree, the operation of decision rules etc.