一种改进的决策树分类属性选择方法

Improved classification attribute selection scheme for decision tree

分析了ID3算法的基本原理、实现步骤及现有两种改进分类算法的优缺点,针对ID3算法的取值偏向问题和现有两种改进算法在分类时间、分类精确度方面存在的不足,提出了一种新的分类属性选择方案,并利用数学知识对其进行了优化。经实验证明,优化后的方案克服了ID3算法的取值偏向问题,同时在分类时间及分类精确度方面优于ID3算法及现有两种改进的分类算法。

Analyze the basic principles and implementation steps of ID3 and point out the advantages and disadvantages of two existing improved classification algorithms.With the shortcoming of inclining to choose attributes having many values for ID3 and the deficiencies of classification time and classification accuracy for existing two improved classification algorithms,a new attribute selection scheme is proposed and optimized with mathematical knowledge.Experiment results show that the optimized scheme can overcome the above disadvantage of ID3 and has the advantages of classification time and classification accuracy over the existing two classification algorithms.