基于粗糙集的偏最小二乘回归方法  

Method of Partial Least Squares Regression Based on Rough Set

作　　者：张小海[1] 金家善[1] 耿俊豹[1] 陈国卫[1] 孙林凯[1] 张军[1] 

机构地区：[1]海军工程大学船舶与动力学院,武汉430033

出　　处：《上海交通大学学报》2010年第12期1678-1681,1686,共5页Journal of Shanghai Jiaotong University

基　　金：中国博士后科研基金资助项目(20080431380)

摘　　要：针对变量之间多重相关性导致最小二乘估计失效的问题,提出基于粗糙集改进偏最小二乘回归建模方法.首先,利用粗糙集对数据进行一般约简,去除冗余信息,再进行偏最小二乘回归分析,建立回归模型.通过实例计算,并与PLSR、PCR进行比较分析.结果表明:用粗糙集改进的PLSR建模精度为3.65%,分别高于PLSR(4.07%)和PCR(4.45%),从而验证了所提出方法的通用性及实用价值.The method of improved partial least squares regression（PLSR） with rough set was proposed to overcome the problem of multiple-relativity which leads least squares invalidation.The data attribute was reduced and redundant information was deleted by the rough set,then the partial least squares regression was analyzed.The new approach can overcome the effect of the disturbed data when filtering principal components in the independent variables.Comparing with the method of principal component regression（PCR） and PLSR,the precision of the improved PLSR with rough set is 3.65%,which is much higher than that of PLSR with 4.07% and PCR with 4.45% respectively.

关 键 词：粗糙集 偏最小二乘回归 主成分回归 

分 类 号：F55[经济管理—产业经济]