基于一种新的属性选择标准的ID3改进算法

结合ID3算法的不足,提出一种基于属性重要度简化标准的ID3改进算法:a)简化ID3算法的信息熵从而降低算法的计算时间;b)引入属性重要度概念来弥补ID3算法属性选择标准的不足;c)综合a)和b)来实现新的属性选择标准即属性重要度简化标准。在开源的Weka数据挖掘软件环境下进行仿真实验,结果表明该改进算法是可行的,并且在算法的计算时间和准确度方面都优于ID3算法,尤其是在数据样本集规模达到一定数量时,效果更加明显。

Improved ID3 algorithm based on new attributes selection criterion

Combining the defects of the ID3 algorithms, this paper proposed an improved ID3 algorithm based on importance of attributes simplified criterion. This algorithom asimplified the information entropy of ID3 algorithm to reduce the computational time, bintroduced the importance of attribute concept to remedy the inadequacy of the attribute selection criterion of ID3 algorithm, ccombined in a and b to achieve a new attribute selection criterion which was the important degree of attributes reduction criterion. Simulation experiments were carried out in the open source Weka data mining software. The results show that the improved algorithm is feasible, and the computation time and accuracy are better than the ID3 algorithm, especially in the sample data set to achieve a certain amount, effect is more obvious.