Is there a way to specify the folds (using my data) with H2O AutoML?

Question

I'm trying to specify the folds for cross-validation in H20. I want to select the subset from the data to be a specific fold. For example, fold number 1 corresponds to my subject 1 data. The fold number 2 to my subject 2 and so on. Is there a way to do it?

I checked the docs for H20 but I didn't find a way for specifying the fold based on my data.

Answer 1

Yes, there's a way to do this in H2O AutoML and all the supervised H2O algorithms.

There's an argument called fold_column in which you specify the name of the column that defines the fold ID. So if you have a "subject_id" column, just set fold_column = "subject_id".

More info in the docs.