摘要
多重共线性是多元线性回归分析中的一个重要问题,消除共线性的危害一直是回归分析的一个重点。就此问题介绍了一种Lasso方法,并设计了一种选择最佳模型的方法。通过实例分析,将其与常用方法进行比较,从结果可看出,Lasso回归在处理多重共线性问题上较其他方法更加有效。
High-dimensional multi-linearity has been a very important problem and how to eliminate the multi-linearity hazards regression analysis has been a priority.To address this problem we introduce more popular Lasso method and design a method of selecting best model.A real example is given to illustrate the calculation steps of the Lasso regression and it is compared with commonly used methods.From the results we can see that the Lasso regression is more effective in handling high-dimensional collinear problem when comparing with other methods.
出处
《江南大学学报（自然科学版）》
CAS
2012年第1期87-90,共4页
Joural of Jiangnan University (Natural Science Edition)
基金
国家自然科学基金项目(50979029)
河海大学自然科学基金项目(2008431111)
关键词
Lasso回归
主成分回归
岭回归
最小角回归算法
AIC准则
BIC准则
Lasso regression
principal component regression
ridge regression
least angle regression algorithms
AIC criterion
BIC criterion
