动态决策树算法研究

Researches on Dynamic Algorithm of Decision Trees

该文在增量决策树算法的基础上,提出一种能够处理变化数据集的减量决策树算法,提出并证明了减量决策树算法中的三个基本定理,保证了减量决策树算法的可靠性。同时将传统的增量决策树算法与该文所提出的减量决策树算法相结合,构造出一种动态决策树算法,该算法很好地解决了发生增减变化的动态数据集构造决策树的问题,另外动态决策树算法的提出也促进了在线规则提取的发展与完善。

Based on the incremental algorithm,this paper proposes a new algorithm that can induce decision trees from the decreasing datasets.At the same time ,this paper also presents and testifies the three theorems that ensure the valid-ity of the decreasing algorithm.Furthermore,integrating the traditional incremental algorithm with decreasing algorithm,this paper proposes a dynamic algorithm of decision trees,which can extract rules of decision trees from the changeable datasets.The dynamic algorithm can promote the development and perfection of the On-Line Rules Extraction.