Hi everyone,
I wanted to comment that I found question 2 of the questionnaire confusing after the emphasis is put on the need to apply cross validation when analyzing generalisation performance. I decided not to mark options a) and b) because the way they are written it seems that they are done once (as if it were a k-folding of k=1), and not marking them made me fail that point.
greetings 
Lucia
Generalization performance can be obtained either with a train-test split or cross-validation. Indeed, a train-test split is equivalent to a (K=1)-Fold cross-validation. The difference is that (K>1)-Fold cross-validation will give additional information on the stability of such scores through different folds (see for instance the video validation of a model).
Per documentation, we will refer to generalization performance of a model when referring to the test score or test error obtained by comparing the prediction of a model and the true targets.
so there will be two answers. other than cross_validate.