Continuing with the artificial data from Exercise (8.3), graph the crossvalidation function CV((s) and generalized cross-validation function GCV(s) as a function of span, letting the span range...

Continuing with the artificial data from Exercise (8.3), graph the crossvalidation function CV((s) and generalized cross-validation function GCV(s) as a function of span, letting the span range between s ¼ 0:05 and s = 0:95.

(a) Compare the shape of CV(s) with the average squared error ASE(s) function from the preceding exercise. Are the shapes similar?

(b) Now compare the level of CV(s) to that of ASE(s). Are the levels different? Why?

(c) Does GCV(s) do a good job of approximating CV(s)?

(d) Do CV(s) and GCV(s) provide useful guidance for selecting the span in this problem?