Remove old dask-glm based logistic regression

conda/environments/all_cuda-118_arch-x86_64.yaml

@@ -50,7 +50,6 @@ dependencies:
 - numpydoc
 - nvcc_linux-64=11.8
 - packaging
-- pip
 - pydata-sphinx-theme!=0.14.2
 - pylibraft==24.10.*,>=0.0.0a0
 - pynndescent
@@ -78,6 +77,4 @@ dependencies:
 - sysroot_linux-64==2.17
 - treelite==4.3.0
 - umap-learn==0.5.6
-- pip:
-  - dask-glm==0.3.0
 name: all_cuda-118_arch-x86_64

conda/environments/all_cuda-125_arch-x86_64.yaml

@@ -46,7 +46,6 @@ dependencies:
 - numpy>=1.23,<3.0a0
 - numpydoc
 - packaging
-- pip
 - pydata-sphinx-theme!=0.14.2
 - pylibraft==24.10.*,>=0.0.0a0
 - pynndescent
@@ -74,6 +73,4 @@ dependencies:
 - sysroot_linux-64==2.17
 - treelite==4.3.0
 - umap-learn==0.5.6
-- pip:
-  - dask-glm==0.3.0
 name: all_cuda-125_arch-x86_64

dependencies.yaml

@@ -517,15 +517,6 @@ dependencies:
           - statsmodels
           - umap-learn==0.5.6
           - pynndescent
-          - setuptools  # Needed on Python 3.12 for dask-glm, which requires pkg_resources but Python 3.12 doesn't have setuptools by default
-      - output_types: conda
-        packages:
-          - pip
-          - pip:
-              - dask-glm==0.3.0
-      - output_types: pyproject
-        packages:
-          - dask-glm==0.3.0
   test_notebooks:
     common:
       - output_types: [conda, requirements]

python/cuml/cuml/tests/dask/test_dask_logistic_regression.py

@@ -103,92 +103,6 @@ def make_classification_dataset(
     return X, y
 
 
-def select_sk_solver(cuml_solver):
-    if cuml_solver == "newton":
-        return "newton-cg"
-    elif cuml_solver in ["admm", "lbfgs"]:
-        return "lbfgs"
-    else:
-        pytest.xfail("No matched sklearn solver")
-
-
-@pytest.mark.mg
-@pytest.mark.parametrize("nrows", [1e5])
-@pytest.mark.parametrize("ncols", [20])
-@pytest.mark.parametrize("n_parts", [2, 6])
-@pytest.mark.parametrize("fit_intercept", [False, True])
-@pytest.mark.parametrize("datatype", [np.float32, np.float64])
-@pytest.mark.parametrize("gpu_array_input", [False, True])
-@pytest.mark.parametrize(
-    "solver", ["admm", "gradient_descent", "newton", "lbfgs", "proximal_grad"]
-)
-def test_lr_fit_predict_score(
-    nrows,
-    ncols,
-    n_parts,
-    fit_intercept,
-    datatype,
-    gpu_array_input,
-    solver,
-    client,
-):
-    sk_solver = select_sk_solver(cuml_solver=solver)
-
-    def imp():
-        import cuml.comm.serialize  # NOQA
-
-    client.run(imp)
-
-    from cuml.dask.extended.linear_model import (
-        LogisticRegression as cumlLR_dask,
-    )
-
-    n_info = 5
-    nrows = int(nrows)
-    ncols = int(ncols)
-    X, y = make_classification_dataset(datatype, nrows, ncols, n_info)
-
-    gX, gy = _prep_training_data(client, X, y, n_parts)
-
-    if gpu_array_input:
-        gX = gX.values
-        gX._meta = cp.asarray(gX._meta)
-        gy = gy.values
-        gy._meta = cp.asarray(gy._meta)
-
-    cuml_model = cumlLR_dask(
-        fit_intercept=fit_intercept, solver=solver, max_iter=10
-    )
-
-    # test fit and predict
-    cuml_model.fit(gX, gy)
-    cu_preds = cuml_model.predict(gX)
-    accuracy_cuml = accuracy_score(y, cu_preds.compute().get())
-
-    sk_model = skLR(fit_intercept=fit_intercept, solver=sk_solver, max_iter=10)
-    sk_model.fit(X, y)
-    sk_preds = sk_model.predict(X)
-    accuracy_sk = accuracy_score(y, sk_preds)
-
-    assert (accuracy_cuml >= accuracy_sk) | (
-        np.abs(accuracy_cuml - accuracy_sk) < 1e-3
-    )
-
-    # score
-    accuracy_cuml = cuml_model.score(gX, gy).compute().item()
-    accuracy_sk = sk_model.score(X, y)
-
-    assert (accuracy_cuml >= accuracy_sk) | (
-        np.abs(accuracy_cuml - accuracy_sk) < 1e-3
-    )
-
-    # predicted probabilities should differ by <= 5%
-    # even with different solvers (arbitrary)
-    probs_cuml = cuml_model.predict_proba(gX).compute()
-    probs_sk = sk_model.predict_proba(X)[:, 1]
-    assert np.abs(probs_sk - probs_cuml.get()).max() <= 0.05
-
-
 @pytest.mark.mg
 @pytest.mark.parametrize("n_parts", [2])
 @pytest.mark.parametrize("datatype", [np.float32, np.float64])