基于贡献因子的改进决策树属性选择方法

对传统ID3算法的信息熵作近似变换达到简化计算的目的,并针对ID3算法倾向于选择取值较多的属性和排斥属性值分布均匀的属性两个缺点,引入贡献因子来改进ID3算法,使属性选择方法平衡的选择划分属性.实验表明,优化后的属性选择方法降低了分类时间,优化了分类结果并能很好地反映实际情况.

Improved Decision Tree Attribute Selection Method Based on the Contribution Factor

In this paper,the calculation of information entropy of the traditional ID3 algorithm by approximate transformation is simplified.The contribution factor is introduced in this paper to improve the algorithm,aiming at two shortcuts of ID3 algorithm,that is,tending to select more value attributes and have exclusion of attributes whose values are uniformly distributed.It allows the attribute selection method to select division attributes in balance.Through experiment testing,the improved attribute selection method can reduce the classification time,optimize the classification results and reflect the actual situation.