Hi,
I've been reading about the problem of shrinkage. It's driving me mad. Can
someone explain to me what's the point of a shrinkage estimator? In
particular, do we ever need it when we're not using stepwise regression?
Apparently, shrinkage applies even when we're not doing stepwise, but on
the other hand, the OLS estimator is always unbiased. So why on earth do
we need to shrink our OLS estimator?
Tim
|