一种基于粗糙集的决策树生成算法

An Algorithm for Decision Tree Construction Based on Rough Sets

针对决策树的构造和修剪通常不能同时进行所产生的效率低下的问题,提出了基于粗糙集理论中知识依赖性的决策树构造方法。利用优先策略,将知识依赖性同时作为属性约简和建树的准则,在决策树预修剪的同时进行节点生成,大大提高了决策树构造的效率。使用Fisher's iris数据集对基于粗糙集理论中知识依赖性的决策树生成算法和用回归拟合方法的决策树生成算法进行比较。实验结果表明,前者的分类精度和决策树模型的复杂程度要明显优于后者。

Aimed at the inefficient problem, in which the decision tree＇s construction and pruning can not be dealt with simultaneously, an algorithm based wholly on the knowledge dependence in rough set theory is proposed. With the method, the concept of knowledge dependence is used to execute pre-pruning and nodes selecting for the decision tree simultaneously. Then the efficiency of decision tree＇s construction is greatly improved. Fisher＇s iris data set is used to compare the decision tree construction algorithm based on the knowledge dependence and the regression approach. The experimental results show that the previous method is better in classification precision and classification model complicacy.