粗糙逼近近似度量在分类规则挖掘中的应用

Application of rough approaching approximation measurement to data mining classification rules in data ware

为了提高决策系统的分类质量,探讨了一种在数据仓库中基于粗糙逼近近似度量的挖掘分类规则策略.首先介绍了数据集中挖掘分类规则的基本原理,并利用粗糙集理论中粗糙逼近近似度量概念,根据决策表条件属性重要性度量及条件属性对决策类划分的逼近近似度量,提出了基于改进粗糙逼近近似度量的数据挖掘进行属性约减方法,最后举例说明了如何在数据库中发现分类规则.实验结果表明此方法挖掘出的规则简练且合理可靠.

In order to improve the classification quality of decision system,a strategy of data mining classification rules based on rough approaching approximation measurement in data ware is proposed.The basic principle of data mining classification rules in data sets is firstly introduced.Then,a attribute reduction method in data mining based on improved rough approaching approximation measurement is put forward according to the measurement of the importance of the conditional attributes in decision table and the approaching approximation measurement of decision classification through conditional attributes.Finally,a simple example is presented for showing how to find classification rules in data-base,and it is shown that the presented data classification rules are rational and reliable.