[R] Hastie - Tibshirani - Friedman pg 141 & nnet question

[G Ilhamto]

Dear helper,

I am working with a nnet using large data set (23K) and have some questions.
I have a binary response (occurrence & non-occurrence of event) with 8
predictors.

(1) How can I reproduce plot in Hastie et al. (page 141), i.e. natural
cubic splines of tensor product?
(2) How does nnet treat the response. It seems that by default it
treats Y as numeric (?). Can I change the response as factor? or, does
it matter?

Thank you,

Ilham