feature req: add w parameter (ex sample_weight) to progressive_val_score

[niccolopetti]

The fit method of sklearn has a sample_weight parameter to pass weights of the instances, and the learn_one method of river also has that attribute, recently renamed to w in #1457 , currently if we had a list of weights and instances to perform our training using our weights we would have to use the learn_one method manually iterating over our data,
it would be much more comfortable to be able to pass the weights with the instances as it happens with sklearn.

it would be nice to be able to do something like:

from river import datasets
from river import evaluate
from river import metrics

evaluate.progressive_val_score(
    model=model,
    dataset=datasets.Phishing(),
    metric=metrics.ROCAUC(),
    print_every=200,
    w=weights
)

where weights contain for each instance of the dataset the corresponding weight

[MaxHalford]

This is a good point. I think a more elegant approach would be to allow the dataset to yield an optional weight for each instance. I can take care of implementing this, but I don't have any time right now.

[niccolopetti]

This is a good point. I think a more elegant approach would be to allow the dataset to yield an optional weight for each instance. I can take care of implementing this, but I don't have any time right now.

Thanks, I'm glad we agree and I hope to see it implemented in the future

[Ahanmr]

I'm trying to work on this, is this still an issue?

[niccolopetti]

I'm trying to work on this, is this still an issue?

Thanks, it is still an issue indeed