Ex. 6.7

Ex. 6.7

Derive an expression for the leave-one-out cross-validated residual sum-of-squares for local polynomial regression.

Soln. 6.7

Note that local regression smoothers are linear estimators, and we can write

\[\begin{equation} \hat{\mathbf{f}} = \bb{S}_\lambda \bb{y} \non \end{equation}\]

where \(\{\bb{S}_\lambda\}_{ij} = l_i(x_j)\) for \(l_i(x)\) defined by (6.8) in the text. Then by Ex. 7.3 we know

\[\begin{equation} y_i - \hat f^{-i}(x_i) = \frac{y_i - \hat f(x_i)}{1-\{\bb{S}_\lambda\}_{ii}}.\non \end{equation}\]