From e49b3aa04f9ada174f2718e9610dd464d988c148 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Fri, 11 Feb 2022 14:35:03 -0600
Subject: [PATCH 01/18] meat finished but not checked

---
 sklearn/linear_model/_glm/glm.py | 42 +++++++++++++++++++++++---------
 1 file changed, 31 insertions(+), 11 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index d7af8ae60d8b6..f56381e6db980 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -13,7 +13,7 @@
 
 from ...base import BaseEstimator, RegressorMixin
 from ...utils.optimize import _check_optimize_result
-from ...utils import check_scalar
+from ...utils import check_scalar, check_array
 from ...utils.validation import check_is_fitted, _check_sample_weight
 from ..._loss.glm_distribution import (
     ExponentialDispersionModel,
@@ -68,12 +68,16 @@ class GeneralizedLinearRegressor(RegressorMixin, BaseEstimator):
 
     Parameters
     ----------
-    alpha : float, default=1
-        Constant that multiplies the penalty term and thus determines the
+    alpha : {float, iterable}, default=1
+        Constant(s) that multiplies the penalty term and thus determines the
         regularization strength. ``alpha = 0`` is equivalent to unpenalized
         GLMs. In this case, the design matrix `X` must have full column rank
         (no collinearities).
         Values must be in the range `[0.0, inf)`.
+        If alpha is a scalar then the value is applied to all non-intercept terms
+        If alpha is an iterable then each value must be in the range `[0.0, inf)`
+            and the size of the iterable must be equal to the input design matrix.
+            If alpha is greater than 1 dimension it will be converted to 1 dimension.
 
     fit_intercept : bool, default=True
         Specifies if a constant (a.k.a. bias or intercept) should be
@@ -214,13 +218,27 @@ def fit(self, X, y, sample_weight=None):
                     "got (link={0})".format(self.link)
                 )
 
-        check_scalar(
-            self.alpha,
-            name="alpha",
-            target_type=numbers.Real,
-            min_val=0.0,
-            include_boundaries="left",
-        )
+        if hasattr(self.alpha, '__iter__'):
+            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
+            if self.alpha.size != X.shape[1]:
+                raise ValueError(f'Alpha must have length equivalent to input array. Input array is shape {X.shape[1]}'
+                                 f'while alpha is of length {self.alpha.size}')
+            for i, val in enumerate(self.alpha):
+                check_scalar(val
+                             , name=f'alpha index [{i}]'
+                             , target_type=numbers.Real
+                             , min_val=0.
+                             , include_boundaries='left')
+        else:
+            check_scalar(
+                self.alpha,
+                name="alpha",
+                target_type=numbers.Real,
+                min_val=0.0,
+                include_boundaries="left",
+            )
+            self.alpha = np.full(X.shape[1], self.alpha)
+
         if not isinstance(self.fit_intercept, bool):
             raise ValueError(
                 "The argument fit_intercept must be bool; got {0}".format(
@@ -279,7 +297,9 @@ def fit(self, X, y, sample_weight=None):
                     family.__class__.__name__
                 )
             )
-        # TODO: if alpha=0 check that X is not rank deficient
+        # check that matrix is full rank when alpha = 0.
+        if np.max(self.alpha) == 0. and np.linalg.matrix_rank(X) < X.shape[1]:
+            raise ValueError('Design matrix must be full rank when alpha = 0.')
 
         # rescaling of sample_weight
         #

From 6a0f4d4831ccfe0d6aa8b10da13cf9413222244f Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 14 Feb 2022 11:24:38 -0600
Subject: [PATCH 02/18] unit tests fixed and added

---
 sklearn/linear_model/_glm/glm.py            | 44 ++++++++++-----------
 sklearn/linear_model/_glm/tests/test_glm.py | 28 ++++++++++++-
 2 files changed, 47 insertions(+), 25 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index f56381e6db980..f4d8408f14b4a 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -217,28 +217,6 @@ def fit(self, X, y, sample_weight=None):
                     "an element of ['auto', 'identity', 'log']; "
                     "got (link={0})".format(self.link)
                 )
-
-        if hasattr(self.alpha, '__iter__'):
-            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
-            if self.alpha.size != X.shape[1]:
-                raise ValueError(f'Alpha must have length equivalent to input array. Input array is shape {X.shape[1]}'
-                                 f'while alpha is of length {self.alpha.size}')
-            for i, val in enumerate(self.alpha):
-                check_scalar(val
-                             , name=f'alpha index [{i}]'
-                             , target_type=numbers.Real
-                             , min_val=0.
-                             , include_boundaries='left')
-        else:
-            check_scalar(
-                self.alpha,
-                name="alpha",
-                target_type=numbers.Real,
-                min_val=0.0,
-                include_boundaries="left",
-            )
-            self.alpha = np.full(X.shape[1], self.alpha)
-
         if not isinstance(self.fit_intercept, bool):
             raise ValueError(
                 "The argument fit_intercept must be bool; got {0}".format(
@@ -286,7 +264,25 @@ def fit(self, X, y, sample_weight=None):
             y_numeric=True,
             multi_output=False,
         )
-
+        if hasattr(self.alpha, '__iter__') and not isinstance(self.alpha, str):
+            for i, val in enumerate(self.alpha):
+                check_scalar(val
+                             , name=f'alpha at index {i}'
+                             , target_type=numbers.Real
+                             , min_val=0.
+                             , include_boundaries='left')
+            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
+            if self.alpha.size != X.shape[1]:
+                raise ValueError(f'X width is {X.shape[1]} while alpha is of length {self.alpha.size}')
+        else:
+            check_scalar(
+                self.alpha,
+                name="alpha",
+                target_type=numbers.Real,
+                min_val=0.0,
+                include_boundaries="left",
+            )
+            self.alpha = np.full(X.shape[1], self.alpha)
         weights = _check_sample_weight(sample_weight, X)
 
         _, n_features = X.shape
@@ -299,7 +295,7 @@ def fit(self, X, y, sample_weight=None):
             )
         # check that matrix is full rank when alpha = 0.
         if np.max(self.alpha) == 0. and np.linalg.matrix_rank(X) < X.shape[1]:
-            raise ValueError('Design matrix must be full rank when alpha = 0.')
+            raise ValueError('X must be full rank when alpha = 0.')
 
         # rescaling of sample_weight
         #
diff --git a/sklearn/linear_model/_glm/tests/test_glm.py b/sklearn/linear_model/_glm/tests/test_glm.py
index 87fe2b51f4d28..991d82e63b715 100644
--- a/sklearn/linear_model/_glm/tests/test_glm.py
+++ b/sklearn/linear_model/_glm/tests/test_glm.py
@@ -195,6 +195,32 @@ def test_glm_scalar_argument(Estimator, params, err_type, err_msg):
         glm.fit(X, y)
 
 
+@pytest.mark.parametrize(
+    "Estimator",
+    [GeneralizedLinearRegressor, PoissonRegressor, GammaRegressor, TweedieRegressor],
+)
+@pytest.mark.parametrize(
+    "params, err_type, err_msg",
+    [
+        ({"alpha": [1, '2']}, TypeError, "alpha at index 1 must be an instance of float, not str."),
+        ({"alpha": [1, 2, 3], "fit_intercept": True}, ValueError
+         , 'X width is 2 while alpha is of length 3'),
+        ({"alpha": [1, 2, 3], "fit_intercept": False}, ValueError
+         , 'X width is 2 while alpha is of length 3'),
+        ({"alpha": [-2, 2]}, ValueError, "alpha at index 0 == -2, must be >= 0.0"),
+        ({"alpha": [0., 0.]}, ValueError, "X must be full rank when alpha = 0."),
+    ]
+)
+def test_glm_alpha_array(Estimator, params, err_type, err_msg):
+    """Test GLM for invalid alpha input when alpha is an iterable"""
+    X = [[1,2], [2,4]]
+    y = [1, 2]
+    glm = Estimator(**params)
+    with pytest.raises(err_type, match=err_msg):
+        glm.fit(X, y)
+
+
+
 @pytest.mark.parametrize("warm_start", ["not bool", 1, 0, [True]])
 def test_glm_warm_start_argument(warm_start):
     """Test GLM for invalid warm_start argument."""
@@ -494,4 +520,4 @@ def test_tweedie_regression_family(regression_data):
     ],
 )
 def test_tags(estimator, value):
-    assert estimator._get_tags()["requires_positive_y"] is value
+    assert estimator._get_tags()["requires_positive_y"] is value
\ No newline at end of file

From 83dde0e13bcd697e1dac3e24fe7042a89109571f Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Fri, 11 Feb 2022 14:35:03 -0600
Subject: [PATCH 03/18] meat finished but not checked

---
 sklearn/linear_model/_glm/glm.py | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index f4d8408f14b4a..c34378d78320a 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -217,6 +217,28 @@ def fit(self, X, y, sample_weight=None):
                     "an element of ['auto', 'identity', 'log']; "
                     "got (link={0})".format(self.link)
                 )
+
+        if hasattr(self.alpha, '__iter__'):
+            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
+            if self.alpha.size != X.shape[1]:
+                raise ValueError(f'Alpha must have length equivalent to input array. Input array is shape {X.shape[1]}'
+                                 f'while alpha is of length {self.alpha.size}')
+            for i, val in enumerate(self.alpha):
+                check_scalar(val
+                             , name=f'alpha index [{i}]'
+                             , target_type=numbers.Real
+                             , min_val=0.
+                             , include_boundaries='left')
+        else:
+            check_scalar(
+                self.alpha,
+                name="alpha",
+                target_type=numbers.Real,
+                min_val=0.0,
+                include_boundaries="left",
+            )
+            self.alpha = np.full(X.shape[1], self.alpha)
+
         if not isinstance(self.fit_intercept, bool):
             raise ValueError(
                 "The argument fit_intercept must be bool; got {0}".format(
@@ -295,7 +317,11 @@ def fit(self, X, y, sample_weight=None):
             )
         # check that matrix is full rank when alpha = 0.
         if np.max(self.alpha) == 0. and np.linalg.matrix_rank(X) < X.shape[1]:
+<<<<<<< HEAD
             raise ValueError('X must be full rank when alpha = 0.')
+=======
+            raise ValueError('Design matrix must be full rank when alpha = 0.')
+>>>>>>> 39b122d50 (meat finished but not checked)
 
         # rescaling of sample_weight
         #

From 314105bef90e5a71005d9f92e4df929da4fdaef7 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 14 Feb 2022 11:24:38 -0600
Subject: [PATCH 04/18] unit tests fixed and added

---
 sklearn/linear_model/_glm/glm.py | 26 --------------------------
 1 file changed, 26 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index c34378d78320a..f4d8408f14b4a 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -217,28 +217,6 @@ def fit(self, X, y, sample_weight=None):
                     "an element of ['auto', 'identity', 'log']; "
                     "got (link={0})".format(self.link)
                 )
-
-        if hasattr(self.alpha, '__iter__'):
-            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
-            if self.alpha.size != X.shape[1]:
-                raise ValueError(f'Alpha must have length equivalent to input array. Input array is shape {X.shape[1]}'
-                                 f'while alpha is of length {self.alpha.size}')
-            for i, val in enumerate(self.alpha):
-                check_scalar(val
-                             , name=f'alpha index [{i}]'
-                             , target_type=numbers.Real
-                             , min_val=0.
-                             , include_boundaries='left')
-        else:
-            check_scalar(
-                self.alpha,
-                name="alpha",
-                target_type=numbers.Real,
-                min_val=0.0,
-                include_boundaries="left",
-            )
-            self.alpha = np.full(X.shape[1], self.alpha)
-
         if not isinstance(self.fit_intercept, bool):
             raise ValueError(
                 "The argument fit_intercept must be bool; got {0}".format(
@@ -317,11 +295,7 @@ def fit(self, X, y, sample_weight=None):
             )
         # check that matrix is full rank when alpha = 0.
         if np.max(self.alpha) == 0. and np.linalg.matrix_rank(X) < X.shape[1]:
-<<<<<<< HEAD
             raise ValueError('X must be full rank when alpha = 0.')
-=======
-            raise ValueError('Design matrix must be full rank when alpha = 0.')
->>>>>>> 39b122d50 (meat finished but not checked)
 
         # rescaling of sample_weight
         #

From df064c7f2a67110639dfbe61f43cc7d888372366 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 14 Feb 2022 11:40:44 -0600
Subject: [PATCH 05/18] pep8 and black

---
 sklearn/linear_model/_glm/glm.py            | 23 +++++++++-------
 sklearn/linear_model/_glm/tests/test_glm.py | 29 ++++++++++++++-------
 2 files changed, 33 insertions(+), 19 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index f4d8408f14b4a..451d2a045d647 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -264,16 +264,21 @@ def fit(self, X, y, sample_weight=None):
             y_numeric=True,
             multi_output=False,
         )
-        if hasattr(self.alpha, '__iter__') and not isinstance(self.alpha, str):
+        if hasattr(self.alpha, "__iter__") and not isinstance(self.alpha, str):
             for i, val in enumerate(self.alpha):
-                check_scalar(val
-                             , name=f'alpha at index {i}'
-                             , target_type=numbers.Real
-                             , min_val=0.
-                             , include_boundaries='left')
+                check_scalar(
+                    val,
+                    name=f"alpha at index {i}",
+                    target_type=numbers.Real,
+                    min_val=0.0,
+                    include_boundaries="left",
+                )
             self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
             if self.alpha.size != X.shape[1]:
-                raise ValueError(f'X width is {X.shape[1]} while alpha is of length {self.alpha.size}')
+                raise ValueError(
+                    f"X width is {X.shape[1]} while alpha is of length"
+                    f" {self.alpha.size}"
+                )
         else:
             check_scalar(
                 self.alpha,
@@ -294,8 +299,8 @@ def fit(self, X, y, sample_weight=None):
                 )
             )
         # check that matrix is full rank when alpha = 0.
-        if np.max(self.alpha) == 0. and np.linalg.matrix_rank(X) < X.shape[1]:
-            raise ValueError('X must be full rank when alpha = 0.')
+        if np.max(self.alpha) == 0.0 and np.linalg.matrix_rank(X) < X.shape[1]:
+            raise ValueError("X must be full rank when alpha = 0.")
 
         # rescaling of sample_weight
         #
diff --git a/sklearn/linear_model/_glm/tests/test_glm.py b/sklearn/linear_model/_glm/tests/test_glm.py
index 991d82e63b715..f87d545a8e4e7 100644
--- a/sklearn/linear_model/_glm/tests/test_glm.py
+++ b/sklearn/linear_model/_glm/tests/test_glm.py
@@ -202,25 +202,34 @@ def test_glm_scalar_argument(Estimator, params, err_type, err_msg):
 @pytest.mark.parametrize(
     "params, err_type, err_msg",
     [
-        ({"alpha": [1, '2']}, TypeError, "alpha at index 1 must be an instance of float, not str."),
-        ({"alpha": [1, 2, 3], "fit_intercept": True}, ValueError
-         , 'X width is 2 while alpha is of length 3'),
-        ({"alpha": [1, 2, 3], "fit_intercept": False}, ValueError
-         , 'X width is 2 while alpha is of length 3'),
+        (
+            {"alpha": [1, "2"]},
+            TypeError,
+            "alpha at index 1 must be an instance of float, not str.",
+        ),
+        (
+            {"alpha": [1, 2, 3], "fit_intercept": True},
+            ValueError,
+            "X width is 2 while alpha is of length 3",
+        ),
+        (
+            {"alpha": [1, 2, 3], "fit_intercept": False},
+            ValueError,
+            "X width is 2 while alpha is of length 3",
+        ),
         ({"alpha": [-2, 2]}, ValueError, "alpha at index 0 == -2, must be >= 0.0"),
-        ({"alpha": [0., 0.]}, ValueError, "X must be full rank when alpha = 0."),
-    ]
+        ({"alpha": [0.0, 0.0]}, ValueError, "X must be full rank when alpha = 0."),
+    ],
 )
 def test_glm_alpha_array(Estimator, params, err_type, err_msg):
     """Test GLM for invalid alpha input when alpha is an iterable"""
-    X = [[1,2], [2,4]]
+    X = [[1, 2], [2, 4]]
     y = [1, 2]
     glm = Estimator(**params)
     with pytest.raises(err_type, match=err_msg):
         glm.fit(X, y)
 
 
-
 @pytest.mark.parametrize("warm_start", ["not bool", 1, 0, [True]])
 def test_glm_warm_start_argument(warm_start):
     """Test GLM for invalid warm_start argument."""
@@ -520,4 +529,4 @@ def test_tweedie_regression_family(regression_data):
     ],
 )
 def test_tags(estimator, value):
-    assert estimator._get_tags()["requires_positive_y"] is value
\ No newline at end of file
+    assert estimator._get_tags()["requires_positive_y"] is value

From 223e9bb8e0322fcedecb3af99f0ae5b65386d756 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 15 Feb 2022 08:54:06 -0600
Subject: [PATCH 06/18] circleCI test and undid transformation of scalar alpha
 to ndarray

---
 sklearn/linear_model/_glm/glm.py | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index 451d2a045d647..171ffb31555b6 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -13,7 +13,7 @@
 
 from ...base import BaseEstimator, RegressorMixin
 from ...utils.optimize import _check_optimize_result
-from ...utils import check_scalar, check_array
+from ...utils import check_scalar
 from ...utils.validation import check_is_fitted, _check_sample_weight
 from ..._loss.glm_distribution import (
     ExponentialDispersionModel,
@@ -287,7 +287,6 @@ def fit(self, X, y, sample_weight=None):
                 min_val=0.0,
                 include_boundaries="left",
             )
-            self.alpha = np.full(X.shape[1], self.alpha)
         weights = _check_sample_weight(sample_weight, X)
 
         _, n_features = X.shape

From 42903bafcfe5fb198e63375aee438294cb04104b Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 15 Feb 2022 15:47:19 -0600
Subject: [PATCH 07/18] updated doc whats_new

---
 doc/whats_new/_contributors.rst |  4 +++-
 doc/whats_new/v1.1.rst          | 37 +++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/doc/whats_new/_contributors.rst b/doc/whats_new/_contributors.rst
index ca0f8ede93afa..f7b44760f5ab9 100644
--- a/doc/whats_new/_contributors.rst
+++ b/doc/whats_new/_contributors.rst
@@ -176,4 +176,6 @@
 
 .. _Nicolas Hug: https://github.com/NicolasHug
 
-.. _Guillaume Lemaitre: https://github.com/glemaitre
\ No newline at end of file
+.. _Guillaume Lemaitre: https://github.com/glemaitre
+
+.. _Xiao Wei: https://github.com/xiaowei1234
\ No newline at end of file
diff --git a/doc/whats_new/v1.1.rst b/doc/whats_new/v1.1.rst
index f9ffc5d329b51..985b0aae7b526 100644
--- a/doc/whats_new/v1.1.rst
+++ b/doc/whats_new/v1.1.rst
@@ -436,6 +436,43 @@ Changelog
   `alpha`, `max_iter` and `tol`.
   :pr:`22240` by :user:`Arturo Amor <ArturoAmorQ>`.
 
+- |Fix| :class:`linear_model.LogisticRegression` and
+  :class:`linear_model.LogisticRegressionCV` now set the `n_iter_` attribute
+  with a shape that respects the docstring and that is consistent with the shape
+  obtained when using the other solvers in the one-vs-rest setting. Previously,
+  it would record only the maximum of the number of iterations for each binary
+  sub-problem while now all of them are recorded. :pr:`21998` by
+  :user:`Olivier Grisel <ogrisel>`.
+
+- |Enhancement| :class:`linear_model.BayesianRidge` and
+  :class:`linear_model.ARDRegression` now preserve float32 dtype. :pr:`9087` by
+  :user:`Arthur Imbert <Henley13>` and :pr:`22525` by :user:`Meekail Zain <micky774>`.
+
+:mod:`sklearn.manifold`
+.......................
+
+- |Feature| :class:`sklearn.manifold.Isomap` now supports radius-based
+  neighbors via the `radius` argument.
+  :pr:`19794` by :user:`Zhehao Liu <MaxwellLZH>`.
+
+- |Enhancement| :func:`manifold.spectral_embedding` and
+  :class:`manifold.SpectralEmbedding` supports `np.float32` dtype and will
+  preserve this dtype.
+  :pr:`21534` by :user:`Andrew Knyazev <lobpcg>`.
+
+- |Enhancement| Adds `get_feature_names_out` to :class:`manifold.Isomap`
+  and :class:`manifold.LocallyLinearEmbedding`. :pr:`22254` by `Thomas Fan`_.
+
+- |Fix| :func:`manifold.spectral_embedding` now uses Gaussian instead of
+  the previous uniform on [0, 1] random initial approximations to eigenvectors
+  in eigen_solvers `lobpcg` and `amg` to improve their numerical stability.
+  :pr:`21565` by :user:`Andrew Knyazev <lobpcg>`.
+- |Enhancement| :class:`linear_model.GeneralizedLinearRegressor` fit method
+  will now allow penalty term `alpha` to be an iterable of same length as number
+  of features in X in addition to a scalar value. Will also check that X matrix
+  is full rank if penalty term `alpha` is set to 0.0 or array of 0.0s.
+  :pr:`11566` by :user:`Xiao Wei <xiaowei1234>`.
+
 :mod:`sklearn.metrics`
 ......................
 

From 7062825096bb268349d77c17158c2a3ef76474e0 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 15 Feb 2022 15:52:07 -0600
Subject: [PATCH 08/18] wrong PR #

---
 doc/whats_new/v1.1.rst | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/doc/whats_new/v1.1.rst b/doc/whats_new/v1.1.rst
index 985b0aae7b526..d6f97b3f0377a 100644
--- a/doc/whats_new/v1.1.rst
+++ b/doc/whats_new/v1.1.rst
@@ -471,7 +471,7 @@ Changelog
   will now allow penalty term `alpha` to be an iterable of same length as number
   of features in X in addition to a scalar value. Will also check that X matrix
   is full rank if penalty term `alpha` is set to 0.0 or array of 0.0s.
-  :pr:`11566` by :user:`Xiao Wei <xiaowei1234>`.
+  :pr:`22485` by :user:`Xiao Wei <xiaowei1234>`.
 
 :mod:`sklearn.metrics`
 ......................

From 01141d8da1f228086319838e93639c6e16d20d3b Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 22 Feb 2022 08:34:14 -0600
Subject: [PATCH 09/18] updated docstring

---
 sklearn/linear_model/_glm/glm.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index 171ffb31555b6..d8fb67156814c 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -68,16 +68,16 @@ class GeneralizedLinearRegressor(RegressorMixin, BaseEstimator):
 
     Parameters
     ----------
-    alpha : {float, iterable}, default=1
+    alpha : {float, array-like} if array like then of size n_features, default=1
         Constant(s) that multiplies the penalty term and thus determines the
         regularization strength. ``alpha = 0`` is equivalent to unpenalized
         GLMs. In this case, the design matrix `X` must have full column rank
         (no collinearities).
         Values must be in the range `[0.0, inf)`.
         If alpha is a scalar then the value is applied to all non-intercept terms
-        If alpha is an iterable then each value must be in the range `[0.0, inf)`
-            and the size of the iterable must be equal to the input design matrix.
-            If alpha is greater than 1 dimension it will be converted to 1 dimension.
+        If alpha is an array-like then each value must be in the range `[0.0, inf)`
+            and the length must equal to n_features.
+        If alpha is greater than 1 dimension it will be converted to 1 dimension.
 
     fit_intercept : bool, default=True
         Specifies if a constant (a.k.a. bias or intercept) should be

From 079cfad97c077a35b40f318dd238be4112ae9ce0 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 1 Mar 2022 08:56:27 -0600
Subject: [PATCH 10/18] Update doc/whats_new/v1.1.rst

Co-authored-by: Alexandre Gramfort <alexandre.gramfort@m4x.org>
---
 doc/whats_new/v1.1.rst | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/doc/whats_new/v1.1.rst b/doc/whats_new/v1.1.rst
index d6f97b3f0377a..99936a8620d64 100644
--- a/doc/whats_new/v1.1.rst
+++ b/doc/whats_new/v1.1.rst
@@ -436,6 +436,13 @@ Changelog
   `alpha`, `max_iter` and `tol`.
   :pr:`22240` by :user:`Arturo Amor <ArturoAmorQ>`.
 
+
+- |Enhancement| :class:`linear_model.GeneralizedLinearRegressor` fit method
+  will now allow penalty term `alpha` to be an iterable of same length as the number
+  of features in X in addition to a scalar value. Will also check that X matrix
+  is full rank if penalty term `alpha` is set to 0.0 or array of 0.0s.
+  :pr:`22485` by :user:`Xiao Wei <xiaowei1234>`.
+
 - |Fix| :class:`linear_model.LogisticRegression` and
   :class:`linear_model.LogisticRegressionCV` now set the `n_iter_` attribute
   with a shape that respects the docstring and that is consistent with the shape

From 038fbd8c171fb85ea968f822ababef04aa3a9302 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 1 Mar 2022 08:56:39 -0600
Subject: [PATCH 11/18] Update sklearn/linear_model/_glm/glm.py

Co-authored-by: Alexandre Gramfort <alexandre.gramfort@m4x.org>
---
 sklearn/linear_model/_glm/glm.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index d8fb67156814c..bec08ae1c4bdd 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -68,7 +68,7 @@ class GeneralizedLinearRegressor(RegressorMixin, BaseEstimator):
 
     Parameters
     ----------
-    alpha : {float, array-like} if array like then of size n_features, default=1
+    alpha : {float, array-like}, default=1
         Constant(s) that multiplies the penalty term and thus determines the
         regularization strength. ``alpha = 0`` is equivalent to unpenalized
         GLMs. In this case, the design matrix `X` must have full column rank

From c0e0a37cef93845ad5a6c8a28eff945a1ebb99dd Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 1 Mar 2022 08:56:46 -0600
Subject: [PATCH 12/18] Update sklearn/linear_model/_glm/glm.py

Co-authored-by: Alexandre Gramfort <alexandre.gramfort@m4x.org>
---
 sklearn/linear_model/_glm/glm.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index bec08ae1c4bdd..de4eeb1f8059c 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -76,7 +76,7 @@ class GeneralizedLinearRegressor(RegressorMixin, BaseEstimator):
         Values must be in the range `[0.0, inf)`.
         If alpha is a scalar then the value is applied to all non-intercept terms
         If alpha is an array-like then each value must be in the range `[0.0, inf)`
-            and the length must equal to n_features.
+        and the length must equal to n_features.
         If alpha is greater than 1 dimension it will be converted to 1 dimension.
 
     fit_intercept : bool, default=True

From a62c3bf1f6247d2c85f3e60b4eadf4fa10f48b7b Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 1 Mar 2022 09:51:01 -0600
Subject: [PATCH 13/18] self.alpha check revision

---
 sklearn/linear_model/_glm/glm.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index de4eeb1f8059c..0fdbed9612e62 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -7,6 +7,7 @@
 # License: BSD 3 clause
 
 import numbers
+from collections.abc import Iterable
 
 import numpy as np
 import scipy.optimize
@@ -264,7 +265,7 @@ def fit(self, X, y, sample_weight=None):
             y_numeric=True,
             multi_output=False,
         )
-        if hasattr(self.alpha, "__iter__") and not isinstance(self.alpha, str):
+        if isinstance(self.alpha, Iterable):
             for i, val in enumerate(self.alpha):
                 check_scalar(
                     val,
@@ -273,7 +274,7 @@ def fit(self, X, y, sample_weight=None):
                     min_val=0.0,
                     include_boundaries="left",
                 )
-            self.alpha = np.asarray(self.alpha, dtype=np.float_).ravel()
+            self.alpha = np.asarray(self.alpha, dtype=np.float64).ravel()
             if self.alpha.size != X.shape[1]:
                 raise ValueError(
                     f"X width is {X.shape[1]} while alpha is of length"

From 801166ca8288929580fd42f54091986bb349e846 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Tue, 1 Mar 2022 10:27:54 -0600
Subject: [PATCH 14/18] rebuilt and fixed unit tests

---
 sklearn/linear_model/_glm/glm.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index 0fdbed9612e62..88b74493b857c 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -265,7 +265,7 @@ def fit(self, X, y, sample_weight=None):
             y_numeric=True,
             multi_output=False,
         )
-        if isinstance(self.alpha, Iterable):
+        if isinstance(self.alpha, Iterable) and not isinstance(self.alpha, str):
             for i, val in enumerate(self.alpha):
                 check_scalar(
                     val,

From 03f1bc3f14dee9767fcd3bd68bf1d7e02f965645 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Sun, 6 Mar 2022 16:24:02 -0600
Subject: [PATCH 15/18] ENH unit test for glm alpha array

---
 sklearn/linear_model/_glm/tests/test_glm.py | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/sklearn/linear_model/_glm/tests/test_glm.py b/sklearn/linear_model/_glm/tests/test_glm.py
index f87d545a8e4e7..327d4a50e76c0 100644
--- a/sklearn/linear_model/_glm/tests/test_glm.py
+++ b/sklearn/linear_model/_glm/tests/test_glm.py
@@ -230,6 +230,24 @@ def test_glm_alpha_array(Estimator, params, err_type, err_msg):
         glm.fit(X, y)
 
 
+@pytest.mark.parametrize(
+    "Estimator",
+    [GeneralizedLinearRegressor, PoissonRegressor, GammaRegressor, TweedieRegressor],
+)
+def test_glm_alpha_array_reg(Estimator):
+    """Test GLM regression when alpha is an array and 2nd column
+    has different alpha than 1st column
+    """
+    X = np.asarray([[1, 2], [1, 3], [1, 4], [1, 3]])
+    y = np.asarray([2, 2, 3, 2])
+    scalar_coefs = Estimator(alpha=1., fit_intercept=False).fit(X, y).coef_
+    X_scaled = X.copy()
+    X_scaled[:, 1] = X_scaled[:, 1] * 2.
+    array_coefs = Estimator(alpha=[1., 4.], fit_intercept=False).fit(X_scaled, y).coef_
+    array_coefs[1] *= 2
+    assert_allclose(scalar_coefs, array_coefs, atol=1e-4)
+
+
 @pytest.mark.parametrize("warm_start", ["not bool", 1, 0, [True]])
 def test_glm_warm_start_argument(warm_start):
     """Test GLM for invalid warm_start argument."""

From 95dad3ff63361ee20737c7a323677487ba39d336 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 7 Mar 2022 09:36:15 -0600
Subject: [PATCH 16/18] ENH reverted matrix rank check and test

---
 sklearn/linear_model/_glm/glm.py            | 4 +---
 sklearn/linear_model/_glm/tests/test_glm.py | 1 -
 2 files changed, 1 insertion(+), 4 deletions(-)

diff --git a/sklearn/linear_model/_glm/glm.py b/sklearn/linear_model/_glm/glm.py
index 88b74493b857c..1bc327b4cde42 100644
--- a/sklearn/linear_model/_glm/glm.py
+++ b/sklearn/linear_model/_glm/glm.py
@@ -298,9 +298,7 @@ def fit(self, X, y, sample_weight=None):
                     family.__class__.__name__
                 )
             )
-        # check that matrix is full rank when alpha = 0.
-        if np.max(self.alpha) == 0.0 and np.linalg.matrix_rank(X) < X.shape[1]:
-            raise ValueError("X must be full rank when alpha = 0.")
+        # TODO: if alpha=0 check that X is not rank deficient
 
         # rescaling of sample_weight
         #
diff --git a/sklearn/linear_model/_glm/tests/test_glm.py b/sklearn/linear_model/_glm/tests/test_glm.py
index 327d4a50e76c0..c300aead14893 100644
--- a/sklearn/linear_model/_glm/tests/test_glm.py
+++ b/sklearn/linear_model/_glm/tests/test_glm.py
@@ -218,7 +218,6 @@ def test_glm_scalar_argument(Estimator, params, err_type, err_msg):
             "X width is 2 while alpha is of length 3",
         ),
         ({"alpha": [-2, 2]}, ValueError, "alpha at index 0 == -2, must be >= 0.0"),
-        ({"alpha": [0.0, 0.0]}, ValueError, "X must be full rank when alpha = 0."),
     ],
 )
 def test_glm_alpha_array(Estimator, params, err_type, err_msg):

From d06826ab953f592447d823205bdfc3d0c774f252 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 7 Mar 2022 09:40:21 -0600
Subject: [PATCH 17/18] ENH doc cleanup for #22485

---
 doc/whats_new/v1.1.rst | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)

diff --git a/doc/whats_new/v1.1.rst b/doc/whats_new/v1.1.rst
index cbd5894a4ea42..65bf1d01ef1a0 100644
--- a/doc/whats_new/v1.1.rst
+++ b/doc/whats_new/v1.1.rst
@@ -537,10 +537,10 @@ Changelog
   `alpha`, `max_iter` and `tol`.
   :pr:`22240` by :user:`Arturo Amor <ArturoAmorQ>`.
 
-- |Enhancement| :class:`linear_model.GeneralizedLinearRegressor` fit method
-  will now allow penalty term `alpha` to be an iterable of same length as the number
-  of features in X in addition to a scalar value. Will also check that X matrix
-  is full rank if penalty term `alpha` is set to 0.0 or array of 0.0s.
+- |Enhancement| :class:`linear_model.PoissonRegressor`, `linear_model.GammaRegressor`,
+  and `linear_model.TweedieRegressor` fit methods will now allow penalty term `alpha`
+  to be an iterable of same length as the number of features in X in addition to a
+  scalar value.
   :pr:`22485` by :user:`Xiao Wei <xiaowei1234>`.
 
 - |Fix| :class:`linear_model.LogisticRegression` and
@@ -575,12 +575,6 @@ Changelog
   in eigen_solvers `lobpcg` and `amg` to improve their numerical stability.
   :pr:`21565` by :user:`Andrew Knyazev <lobpcg>`.
 
-- |Enhancement| :class:`linear_model.GeneralizedLinearRegressor` fit method
-  will now allow penalty term `alpha` to be an iterable of same length as number
-  of features in X in addition to a scalar value. Will also check that X matrix
-  is full rank if penalty term `alpha` is set to 0.0 or array of 0.0s.
-  :pr:`22485` by :user:`Xiao Wei <xiaowei1234>`.
-
 - |Enhancement| added `metric_params` to :class:`manifold.TSNE` constructor for
   additional parameters of distance metric to use in optimization.
   :pr:`21805` by :user:`Jeanne Dionisi <jeannedionisi>` and :pr:`22685` by

From 98f820e7574584b45084e772cca0084db0f61231 Mon Sep 17 00:00:00 2001
From: Xiao Wei <xiaoswei@gmail.com>
Date: Mon, 7 Mar 2022 09:56:57 -0600
Subject: [PATCH 18/18] ENH #22485 black

---
 sklearn/linear_model/_glm/tests/test_glm.py | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/sklearn/linear_model/_glm/tests/test_glm.py b/sklearn/linear_model/_glm/tests/test_glm.py
index c300aead14893..8ae4f5e629d84 100644
--- a/sklearn/linear_model/_glm/tests/test_glm.py
+++ b/sklearn/linear_model/_glm/tests/test_glm.py
@@ -239,11 +239,13 @@ def test_glm_alpha_array_reg(Estimator):
     """
     X = np.asarray([[1, 2], [1, 3], [1, 4], [1, 3]])
     y = np.asarray([2, 2, 3, 2])
-    scalar_coefs = Estimator(alpha=1., fit_intercept=False).fit(X, y).coef_
+    scalar_coefs = Estimator(alpha=1.0, fit_intercept=False).fit(X, y).coef_
     X_scaled = X.copy()
-    X_scaled[:, 1] = X_scaled[:, 1] * 2.
-    array_coefs = Estimator(alpha=[1., 4.], fit_intercept=False).fit(X_scaled, y).coef_
-    array_coefs[1] *= 2
+    X_scaled[:, 1] = X_scaled[:, 1] * 2.0
+    array_coefs = (
+        Estimator(alpha=[1.0, 4.0], fit_intercept=False).fit(X_scaled, y).coef_
+    )
+    array_coefs[1] *= 2.0
     assert_allclose(scalar_coefs, array_coefs, atol=1e-4)