Cross validation: Each sample or set left out once?

[corneliusroemer]

I wasn't quite sure whether I understood this sentence correctly:

Gradient boosting machines were applied with 10-fold cross-validation such that every sample was left out once.

If I understand it correclty, you split all samples into ten sets, and leave each set out once, not each sample.

The sentence above appears twice, in captions 1F and 2A.

[gregpoore]

@corneliusroemer

Sorry for the confusion. Both are correct, but my wording was not the clearest:

• The cross-validation uses 10 folds, or ten train-test splits (each 90% train, 10% test).
• A sample is included in only one of the ten 10% test sets, meaning it only has one prediction assigned to it during the 10-fold cross-validation procedure. This is what I meant by each sample was "left out" only once. However, it is not the clearest language, so I've removed that phrase in captions 1F and 2A.

Thanks for question!