[dask] remove extra 'client' kwarg in DaskLGBMRegressor

[jameslamb]

Looks like I missed removing the client argument in DaskLGBMRegressor in #3883 , sorry!

I think in the future, we might be able to use mypy to catch API differences between the Dask package and its scikit-learn equivalents. For now, this PR removes that argument.

[StrikerRUS]

I think in the future, we might be able to use mypy to catch API differences between the Dask package and its scikit-learn equivalents. For now, this PR removes that argument.

I believe this approach you used for init with inspect is more reliable.

LightGBM/tests/python_package_test/test_dask.py

Line 816 in b1e000c

def test_dask_classes_and_sklearn_equivalents_have_identical_constructors_except_client_arg(classes):

[jameslamb]

I believe this approach you used for init with inspect is more reliable.

It is, but at the expense of a bit more maintenance burden. Because the test will have to explicitly articulate which keyword args are currently in the signature, and which are passed through with **kwargs**.

Anyway, for now the inspect() thing is probably good to do. I'll open a PR in a bit with tests like that for fit() and predict().

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.