Meta-Estimators with Multiple Models

What’s a good way to try multiple model types with the meta-estimators (like RandomizedSearchCV) in dask-ml? map?

@mattalhonte-srm Welcome to Discourse!

Could you please share some more details (maybe pseudocode, scikit-learn code) of your intended workflow?

Something to the effect of

with worker_client() as client:
    clf = dcv.RandomizedSearchCV(
        model,
        parameters,
        n_iter=10,
        scheduler=client,
        scoring="f1",
        refit=False,
        return_train_score=True,
    )
    clf.fit(
        train_x,
        loadedtrain_y,
    )

With multiple model types (so like, XGBoost and RandomForest).
Right now I’m just wrapping the above in different Prefect tasks (running on an ephemeral Dask cluster).