基于谱系聚类的粗糙集数据挖掘预处理方法  

Preprocessing algorithm based on pedigree cluster for rough set data mining

作　　者：韩中华[1] 马斌[1] 许可[1] 李宏亮[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,沈阳110168

出　　处：《计算机工程与应用》2008年第2期194-196,共3页Computer Engineering and Applications

基　　金：科技部国际合作重点项目基金(No.2003DF020009)。

摘　　要：介绍了一种基于统计分析的数据离散化方法——谱系聚类法,以胶合板缺陷检测数据为应用对象进行了基于谱系聚类的数据离散化研究,并与其它离散化方法进行了对比分析,对比结果表明经谱系聚类方法离散化后的数据,再进行粗糙集约简时,会有更多的冗余属性和记录被约掉,从而可以降低模型的复杂程度,加快获取知识的进程,提高分类的准确率。工程实践证明谱系聚类是一种有效的可用于数据预处理的离散化方法,结合粗糙集算法可以获取满意的数据挖掘结果。A data decentralize method based on statistical analysis is introduced,which is called pedigree cluster method,the data decentralize research,in which the inspection data of wood veneer are taken as the application,has been done and the comparison between pedigree cluster and several other decentralize method are also made.The comparison result shows that the data handled by pedigree cluster will be deleted more redundant attributes and records after rough sets theory, reduction,the complexity of the model can be reduced,the knowledge acquisition process can be accelerated and the accurate of classification can be improved.It has been proved in engineering practice that pedigree cluster method is an effective data decentralize method used in preprocessing process,and combing with rough set method a satisfied data mining result can be obtained.

关 键 词：粗糙集 离散化 谱系聚类 类平均距离 SAS 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]