Batch Generator using RDataframe

ROOT
1

Hi,

I’ve saw in a post from last year about a batch generator for RDataframe. I’m wondering if this feature is available in the last root version.

If not, there is a practical way to get specific rows from a RDF given their indexes?

Thanks!

2

Dear Micael,

The RBatchGenerator is available in experimental mode in the ROOT releases of the 6.30 series. You can see an example usage here: ROOT: tutorials/tmva/RBatchGenerator_PyTorch.py File Reference

I hope this helps.

Cheers,
D

3

Thanks Danilo!!

Cheers
Micael

4

@Danilo,

do you know if it’s possible to integrate this with StratifiedKFold? I want to create a cross-validation process with some folds and them train the models.

5

Dear Micael,

I add in the loop the ML expert of the CERN SFT group, @moneta

Best,
D

6

Thanks @Danilo!

I want something like this: sklearn.model_selection.StratifiedKFold — scikit-learn 1.4.1 documentation

I need to generate the folds and then use each fold to create the generator for training.

Thanks in advance,
Micael