MASS :: lm.ridge系数

Question

I'm trying to perform ridge regression using the lm.ridge method. My question is how can I get the coefficients of the fitted model? I'm getting different results from calling:

• model$coef
• coef(model)

Which is the correct one? Also, why do I get different results from calling:

• coef(model) and looking at the 1st coefficient, vs.
• coef(model)[1] ?

Answer 1

As ?lm.ridge says (in describing the $coef element of the returned object) [emphasis added]

coef: matrix of coefficients, one row for each value of 'lambda'. Note that these are not on the original scale and are for use by the 'coef' method.

This means, specifically, that the $coef element is not intended for end-users ("if you have to ask ..."). (If you want to see how $coef is translated, inspect MASS:::coef.ridgelm .) In general, it's better practice to use an accessor method such as coef() , when it exists, than to extract components from the guts of a returned object using $ (or @ for S4 objects) - for exactly this reason. Package authors provide coef() methods for a reason ...

I can't replicate your second question. Using the model from ?lm.ridge , the answers seem identical except for the precision with which they are printed ...

> m1 <- lm.ridge(y ~ ., longley)
> coef(m1)
                            GNP    Unemployed  Armed.Forces    Population 
    2946.85636017    0.26352725    0.03648291    0.01116105   -1.73702984 
             Year      Employed 
      -1.41879853    0.23128785 
> coef(m1)[1]       
2946.856