Data augmentation for tabular databunch

I was wondering if anyone has spent some time building ds_tfms that will work on Tabular data.

Basically looking to augment the data as it is being loaded into the batches to help combat over-fitting tabular data.

I believe this should work, in principal, similar to image transformations. For tabular it would be good to be able to specify list of columns to augment.

1 Like

I recommend you look at the imblearn library.