Footnote 35 (chapter 3.8)

[crossxwill]

I think the footnote could be improved, so that non-English speakers do not believe the validation set was sampled from the training and test sets.

As noted in the last paragraph of Section 3.4.7, a better approach would be to allocate a separate data set from the training and test sets and to use this for discovery

Proposed revision:

As noted in the last paragraph of Section 3.4.7, a better approach would be to allocate a data set separate from the training and test sets and to use this for discovery