基于概率测度的数据挖掘扩展模型研究

Data mining expand model research based on probability measure

为了得到数据挖掘过程中分类规则的统计特征,论文提出了一种挖掘概率规则的新方法。首先在经典粗糙集概念的基础上分析概率规则的分类,并将其推广到不确定系统的集合等价关系中,即用条件概率的形式表示研究集合的上下近似空间;然后根据概率规则的测度从条件概率的角度利用条件属性的逼近精度的相关参数进行属性集的约简进而提取分类规则;最后给出了相关的仿真实验结果,结果表明带有概率测度的分类规则更合理。

In order to obtain the statistics characteristic of classification rule of data mining,a new method of mining probability rule is put forward in this paper.Firstly,the classification of probability rule is analyzed on the base of classic rough set concepts and extended to the equal relation of set in the indefinite system,namely,the upper and lower approximation space of research set is expressed in the form of conditional probability ;then, according to the measure of probability rule ,the attributes reduction is carried out and the classification rule is extracted by using the related parameters of condition attributes＇ impend precision from the angle of conditional probability;Finally,the related simulation test result is given and the result shows the classification rules with probability measures is more rational.