diff --git a/dask_ml/linear_model/glm.py b/dask_ml/linear_model/glm.py
index 54abc0685..31ef4542a 100644
--- a/dask_ml/linear_model/glm.py
+++ b/dask_ml/linear_model/glm.py
@@ -38,7 +38,7 @@
 
     C : float
         Regularization strength. Note that ``dask-glm`` solvers use
-        the parameterization :math:`\lambda = 1 / C`
+        the parameterization :math:`\\lambda = 1 / C`
 
     fit_intercept : bool, default True
         Specifies if a constant (a.k.a. bias or intercept) should be
diff --git a/dask_ml/model_selection/_search.py b/dask_ml/model_selection/_search.py
index e38ae384c..1f74052ae 100644
--- a/dask_ml/model_selection/_search.py
+++ b/dask_ml/model_selection/_search.py
@@ -1144,9 +1144,8 @@ def fit(self, X, y=None, groups=None, **fit_params):
         if self.multimetric_:
             if self.refit is not False and (
                 not isinstance(self.refit, str)
-                or
                 # This will work for both dict / list (tuple)
-                self.refit not in scorer
+                or self.refit not in scorer
             ):
                 raise ValueError(
                     "For multi-metric scoring, the parameter "
diff --git a/docs/source/conf.py b/docs/source/conf.py
index 250766eb1..abb7290f1 100644
--- a/docs/source/conf.py
+++ b/docs/source/conf.py
@@ -21,9 +21,10 @@
 import shutil
 import subprocess
 
-# The theme to use for HTML and HTML Help pages.  See the documentation for
-# a list of builtin themes.
+import packaging.version
+
 import dask_sphinx_theme
+from dask_ml import __version__ as version
 
 # import sys
 # sys.path.insert(0, os.path.abspath('.'))
@@ -99,10 +100,8 @@
 # built documents.
 #
 # The short X.Y version.
-version = ""
-# The full version, including alpha/beta/rc tags.
-from dask_ml import __version__ as version
-import packaging.version
+# version = ""
+
 release = packaging.version.parse(version).base_version
 
 # The language for content autogenerated by Sphinx. Refer to documentation
diff --git a/setup.cfg b/setup.cfg
index eabbddcd3..3557a416a 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -22,6 +22,8 @@ include_trailing_comma=True
 force_grid_wrap=0
 combine_as_imports=True
 line_length=88
+skip=
+    docs/source/conf.py
 
 [coverage:run]
 source=dask_ml
diff --git a/tests/compose/test_column_transformer.py b/tests/compose/test_column_transformer.py
index 218ff0992..bf88a74b4 100644
--- a/tests/compose/test_column_transformer.py
+++ b/tests/compose/test_column_transformer.py
@@ -14,13 +14,13 @@
 
 def test_column_transformer():
     a = sklearn.compose.make_column_transformer(
-        (["A"], sklearn.preprocessing.OneHotEncoder(sparse=False)),
-        (["B"], sklearn.preprocessing.StandardScaler()),
+        (sklearn.preprocessing.OneHotEncoder(sparse=False), ["A"]),
+        (sklearn.preprocessing.StandardScaler(), ["B"]),
     )
 
     b = dask_ml.compose.make_column_transformer(
-        (["A"], dask_ml.preprocessing.OneHotEncoder(sparse=False)),
-        (["B"], dask_ml.preprocessing.StandardScaler()),
+        (dask_ml.preprocessing.OneHotEncoder(sparse=False), ["A"]),
+        (dask_ml.preprocessing.StandardScaler(), ["B"]),
     )
 
     a.fit(df)
diff --git a/tests/test_incremental.py b/tests/test_incremental.py
index 072d16684..40fb2bdfc 100644
--- a/tests/test_incremental.py
+++ b/tests/test_incremental.py
@@ -139,7 +139,9 @@ def test_score(xy_classification):
     client = distributed.Client(n_workers=2)
 
     X, y = xy_classification
-    inc = Incremental(SGDClassifier(max_iter=1000, random_state=0), scoring="accuracy")
+    inc = Incremental(
+        SGDClassifier(max_iter=1000, random_state=0, tol=1e-3), scoring="accuracy"
+    )
 
     with client:
         inc.fit(X, y, classes=[0, 1])
diff --git a/tests/test_pca.py b/tests/test_pca.py
index 50aa0dc73..992920aa7 100644
--- a/tests/test_pca.py
+++ b/tests/test_pca.py
@@ -375,8 +375,8 @@ def test_pca_validation():
 
                 assert_raises_regex(
                     ValueError,
-                    "n_components={}L? must be between "
-                    "{}L? and min\(n_samples, n_features\)="
+                    r"n_components={}L? must be between "
+                    "{}L? and min\\(n_samples, n_features\\)="
                     "{}L? with svd_solver='{}'".format(
                         n_components, lower_limit[solver], smallest_d, solver_reported
                     ),
@@ -389,9 +389,9 @@ def test_pca_validation():
 
                 assert_raises_regex(
                     ValueError,
-                    "n_components={}L? must be "
+                    r"n_components={}L? must be "
                     "strictly less than "
-                    "min\(n_samples, n_features\)={}L?"
+                    "min\\(n_samples, n_features\\)={}L?"
                     " with svd_solver='arpack'".format(n_components, smallest_d),
                     dd.PCA(n_components, svd_solver=solver).fit,
                     data,
@@ -689,7 +689,12 @@ def test_pca_bad_solver():
 
 @pytest.mark.parametrize(
     "svd_solver",
-    ["full", pytest.mark.xfail(reason="svd_compressed promotes")("randomized")],
+    [
+        "full",
+        pytest.param(
+            "randomized", marks=pytest.mark.xfail(reason="svd_compressed promotes")
+        ),
+    ],
 )
 def test_pca_float_dtype_preservation(svd_solver):
     # Ensure that PCA does not upscale the dtype when input is float32