基于粒度商的决策树构造算法

Quotient GD based algorithm for design decision tree

以粗糙集理论为基础,结合知识关系具有粒度性质的原理,从条件属性集和决策属性集之间关联度来预测和表达决策属性集的一种优性度量,从而定义了粒度商的概念。基于知识粗糙性的粒度原理,以决策树方法为理论基础,把粒度商的概念应用到决策树方法中,提出了一种新的构建决策树的方法,并详细分析了该算法的优点。实例研究表明,提出的基于粒度商的决策树构造算法是可靠、有效的,为进一步研究知识的粒度计算提供了可行的方法。但没有研究不同粒度世界之间的联系,这方面工作还有待进一步研究。

Take the rough sets theory as the foundation, the union knowledge relations has the granularity nature principle, the degree of association from condition attribute collection and decision attribute collection, it forecasts and the expression decision attribute collection one kind of dominance measure. The quotient GD is defmed fi＇om the attribute connection. Take the decision tree method as the rationale, applies in the decision tree method the granularity business concept, a new method to design the decision tree based on knowledge rough＇ s granularity principle is proposed. The advantages of this method are described in detail. Application practically has been proved that the way of proposed quotient GD based algorithm for design the decision tree is reliable effective. Moreover, it puts forward new methods for further researching granular computing of knowledge. But has no studied the relation between the different granularity world, this aspect works also waits for further studies.