using dask_ml with imblearn

[sephib]

Hi,
I want to try and work on implementing dask_ml when running imblearn.
I've opened an issue on dask_ml issue # 317 however I thought it would be a good idea to ping this repo.
I wanted to ask if someone is already working on such implementation, and if not - is there any additional input that can assist while working on the above issue.

[chkoar]

Do you want to contribute in imbalanced-learn or in dask_ml?

[sephib]

To dask_ml, not sure how this issue can be solved in imbalanced-learn

[glemaitre]

The thing is that we would need to handle dask array and dask dataframe which is not compatible with scikit-learn (at least they will be converted to NumPy array). One need to modify the sampler to handle the dask array natively.

FYI: there is an effort to port the sampler there dask/dask-ml#638 (review)

[vishalvvs]

Any success in having imblearn capabilities in Dask-ml?