Monday, January 10, 2011

Comparing OLS, Ridge Regression, LAR, and LASSO

The following penalized residual sums of squares differentiate Ridge Regression, LAR and LASSO from OLS:


min{e'e + λβ'β)       Ridge Regression

min{e'e + λβ)          Least Angle Regression                                                           

min{e'e + λ|β|)        Least Absolute Shrinkage and Selection Operator       

min(e'e)                   Ordinary Least Squares

No comments:

Post a Comment