首先采用企业的客户数据作为样本数据进行客户的稳定性分析，然后，提出了一种基于ID3算法的改进分类算法，该分类新的算法是在经典ID3算法基础上引入粗糙组合属性的思想，使得期望非叶节点到各叶节点的平均路径最短，从而提升分类的速度和准确率。通过实例对改进算法生成决策树产生的结果分析，表明了该算法生成的决策树结构更简单，时间复杂度更优，算法更有效。%In this paper, we analysed the stability by enterprise customer data as the sample data, and then, we proposed an improved algorithm based on the ID3 algorithm via combining with rough set theory. The new classification algorithm is based on the classical ID3 algorithm to introduce the concept of rough combination attribute, which makes the average path of the expected non leaf nodes to the nodes of the shortest path, so as to improve the speed and accuracy of classification. The structure of decision trees which are constructed by the improved algorithm becomes very exact and simple, the time complexity is better than the ID3 algorithm. Experimental results on the decision tree also show the improved algorithm is effective.
展开▼
