一种新的基于属性频率的属性约简算法  

A Novel Attributes Reduction Algorithm Based on Attributes Frequency

作　　者：龚安[1] 牛秋丽[1] 高博[1] 刘华山[1] 

机构地区：[1]中国石油大学(华东)计算机与通信工程学院,东营257061

出　　处：《微计算机应用》2007年第11期1138-1141,共4页Microcomputer Applications

摘　　要：针对目前粗糙集属性约简速度比较慢、不能得到属性约简集的问题,提出了一种新的属性约简算法。通过理论分析、具体的实例和UCI数据集验证,该算法可以确保得到决策表的一个约简,并能减少计算量,提高计算速度。同时算法中引入了强等价集,很好地解决了属性加权频率值相同的问题。Aimed at the problem of the lower speed of attributes reduction and that can＇ t get the reduction,This paper proposes a novel attributes reduction algorithm. Theoretical analyses, experimental results and UCI dataset show that the algorithm not can get a reduction, but reduces the computing effort and improve the computing efficiency. Meanwhile the strong compressible set has been introduced, so the problem that the values of attributes weight frequency are the same has been solved.

关 键 词：属性约简 属性频率 强等价集 UCI数据集 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]