一种基于有序属性决策系统分类规则提取策略

Classification rule extracting strategy based on decision system with ordered attributes

分类规则的精度取决于分类算法的构造,论文在综合分析基本粗糙集合概念及其约简算法的基础上,阐述了一种基于准则的有序属性决策系统的数据挖掘算法．为此首先介绍了基于有序属性决策系统的集合表达,然后利用有序属性决策系统中准则集与属性集的基本特征构造上下近似扩展模型,得到准则集决策系统的四个相关参数．并进一步提出相应的数据约简与分类规则提取算法。最后给出了用此算法约简有序属性决策系统的算例,实验结果表明此方法挖掘出的规则简练,更具合理性和可靠性。

The precision of classification rule is decided by the construction of classification algorithm. By the concepts and attribute reduction algorithm of basic rough set, a data mining algorithm based on the ordered character of attribute in decision system is proposed in this paper. First, the aggregation expression in decision system with ordered character of attribute is briefly introduced. Then, based on the basic characterization of criteria sets and attribute sets in decision system with ordered attributes, the upper and lower approximation expansion models are constructed to obtain the four relative parameters in decision system with ordered attributes. Thirdly, the corresponding data mining and classification rule extracting algorithm is constructed by using the proposed approach. Finally the rationality of the ordered attribute reduction method is validated by simulation example, and the result shows the rules mined by the method are concise and reliable.