摘要
基于投影寻踪法原理,文章提出了一种将线性回归和目标规划方法相结合的目标投影回归综合评价方法(Target Projection Based on Regression,TPR)。该方法将投影寻踪法、线性回归和目标规划方法结合在一起,引入投影墙的概念,定制了取值范围为[0,1]的标准投影墙,舍去了投影寻踪法中的主观性指标窗口半径R,用xi和yi之间的距离函数取代R,同时还约束了变量系数的标准差、各变量与综合评价结果之间的相关系数以及回归模型的拟合优度。TPR在综合评价过程中可以生成一个回归模型,且回归系数具有权重意义,利用该模型可以对多个评价单位进行对比评价。最后,用两个实例证明了该方法的可行性和合理性,表明TPR可作为一种新的综合评价工具进行推广应用。
Based on the principle of projection pursuit,this paper proposes a comprehensive evaluation method for Target Projection Based on Regression(TPR)that combines linear regression and goal programming methods.This method combines projection pursuit,linear regression,and objective programming methods,introduces the concept of projection walls,customizes standard projection walls with a range of[0,1]values,discards the subjective indicator window radius R in projection pursuit,and replaces it with a distance function between xi and ŷi.At the same time,it also constrains the standard deviation of variable coefficients,the correlation coefficients between each variable and the comprehensive evaluation results,and the goodness of fit of the regression model.The TPR method can generate a regression model in the process of comprehensive evaluation,and the regression coefficient has weight significance.This regression model can be used to conduct comparative evaluation of multiple evaluation units.Finally,two examples are employed to demonstrate the feasibility and rationality of this method,indicating that TPR can be promoted and applied as a new comprehensive evaluation tool.
作者
周晓香
李宝瑜
Zhou Xiaoxiang;Li Baoyu(School of Statistics,Shanxi University of Finance and Economics,Taiyuan 030006,China)
出处
《统计与决策》
北大核心
2026年第7期50-56,共7页
Statistics & Decision
关键词
综合评价
投影寻踪
目标投影回归
目标规划
comprehensive evaluation
projection pursuit
target projection based on regression
goal programming
