Christian Lorentzen
Christian Lorentzen
Regardless of the fact that I'm not an expert for SGD and don't know how well it works for non-smooth objective, this PR would need unit tests similar to https://github.com/scikit-learn/scikit-learn/blob/a5b70b3132467b5e3616178d9ecca6cb7316c400/sklearn/linear_model/tests/test_quantile.py#L153...
> What do maintainers think? Would you first perform such a refactoring? I would prefer such a refactoring. But I'm not 100% sure it is feasible. It would certainly slow...
Update: https://github.com/scikit-learn/scikit-learn/pull/24027 will solve this issue.
@glemaitre Could you merge main so that the doc is rendered again?
I like such a plotting functionality very much. If we add - in a later PR - a binning strategy, this could become the equivalent of the reliability diagram for...
> did you encounter any other types of plots for those specific use cases? `PredictionErrorDisplay` offeres several plots and those go under several names, e.g. - actual vs expected -...
Maybe we can continue as is and introduce a further parameter in the future, like `target_type= "expectation"` or `("quantile", 0.9)`. I think the choice of identification function for generalized residuals...
@bradleyfowler123 Thanks for this PR and your willingness to contribute. Please note that https://github.com/scikit-learn/scikit-learn/issues/21391#issuecomment-962592956 was a suggestion only and not a consensus. From a very quick view, I have to...
We have a lot of requests for new metrics. That's why I opened the issue #21718. IIUC, this PR introduces metrics that are just monotonic mappings of already available metrics:...
What do you mean by "it does not connect to what metrics are in the library"? To know what's wrong so far would help to improve or to make it...