From h2o, is there a way to export N folder cross validation results into a dataframe?

Question

I am using H2O sparking water to built GBM model. I know we can view the N folder cross validation results using code below:

gbm_model.model_performance(xval = True)

But is there a way to save each folder's model performance into a data frame? For example, save each folder's AUC into a data frame.

Answer 1

When creating your model, you can set keep_cross_validation_models=True and then obtain the models by calling cross_validation_models() on your model object. Then you can obtain each model's performance as usual. See the documentation.