在Visual Basic环境下用粗糙集理论约简决策表属性

Reduce the attributes of decision tables with the rough set theory under Visual Basic

决策树是一种简单而且应用广泛的分类算法。一个决策表中包含大量的样本信息,一个样本就代表一条基本的决策规则。为了从决策表中抽取适应性大的规则,就需要对决策表进行属性约简。使用粗糙集理论对决策表进行约简比较出色。在用辨识矩阵方法对决策表进行属性约简时,需要考虑实现约简的一些技术问题。如建立决策表的可辨识矩阵,对所有取值非空集合元素建立相应的析取逻辑表达式,对析取表达式进行合取运算等。在VB编程环境下,实现了该算法的约简过程。实例仿真表明了该算法对于决策表属性约简是可行的。

The decision tree is a simple and widely used classification algorithm. There is a large number of sample information in the decision table. One sample is a decision rule. We need reduce the attributes of the decision tables if we want to draw the adaptable rules. It is remarkable when we reduce the decision table. When using the discernibility matrix method ,various technical issues of the algorithm must be considered, such as building the discernibility matrix of the decision tables, building all of the associated disjunction logical expression of the elements in the non-null set and doing conjunctive operation for the disjunction expression etc. Under VB programming environment,we carry out the reduction process of the algorithm. Examples simulation shows that the method is feasible to the reduction of the attributes of decision.