diff --git a/maint_tools/test_docstrings.py b/maint_tools/test_docstrings.py
index e1a4dd7ca5dd6..687184d6f6500 100644
--- a/maint_tools/test_docstrings.py
+++ b/maint_tools/test_docstrings.py
@@ -34,7 +34,6 @@
     "QuadraticDiscriminantAnalysis",
     "RandomizedSearchCV",
     "RobustScaler",
-    "SGDOneClassSVM",
     "SGDRegressor",
     "SelfTrainingClassifier",
     "SparseRandomProjection",
diff --git a/sklearn/linear_model/_stochastic_gradient.py b/sklearn/linear_model/_stochastic_gradient.py
index b429d8227553a..dd86c9536abd4 100644
--- a/sklearn/linear_model/_stochastic_gradient.py
+++ b/sklearn/linear_model/_stochastic_gradient.py
@@ -1975,8 +1975,8 @@ class SGDOneClassSVM(BaseSGD, OutlierMixin):
         Whether or not the training data should be shuffled after each epoch.
         Defaults to True.
 
-    verbose : integer, optional
-        The verbosity level
+    verbose : int, optional
+        The verbosity level.
 
     random_state : int, RandomState instance or None, optional (default=None)
         The seed of the pseudo random number generator to use when shuffling
@@ -1985,7 +1985,7 @@ class SGDOneClassSVM(BaseSGD, OutlierMixin):
         generator; If None, the random number generator is the RandomState
         instance used by `np.random`.
 
-    learning_rate : string, optional
+    learning_rate : str, optional
         The learning rate schedule to use with `fit`. (If using `partial_fit`,
         learning rate must be controlled directly).
 
@@ -2059,6 +2059,17 @@ class SGDOneClassSVM(BaseSGD, OutlierMixin):
 
         .. versionadded:: 1.0
 
+    See Also
+    --------
+    sklearn.svm.OneClassSVM : Unsupervised Outlier Detection.
+
+    Notes
+    -----
+    This estimator has a linear complexity in the number of training samples
+    and is thus better suited than the `sklearn.svm.OneClassSVM`
+    implementation for datasets with a large number of training samples (say
+    > 10,000).
+
     Examples
     --------
     >>> import numpy as np
@@ -2070,17 +2081,6 @@ class SGDOneClassSVM(BaseSGD, OutlierMixin):
 
     >>> print(clf.predict([[4, 4]]))
     [1]
-
-    See also
-    --------
-    sklearn.svm.OneClassSVM
-
-    Notes
-    -----
-    This estimator has a linear complexity in the number of training samples
-    and is thus better suited than the `sklearn.svm.OneClassSVM`
-    implementation for datasets with a large number of training samples (say
-    > 10,000).
     """
 
     loss_functions = {"hinge": (Hinge, 1.0)}
@@ -2295,6 +2295,8 @@ def partial_fit(self, X, y=None, sample_weight=None):
         ----------
         X : {array-like, sparse matrix}, shape (n_samples, n_features)
             Subset of the training data.
+        y : Ignored
+            Not used, present for API consistency by convention.
 
         sample_weight : array-like, shape (n_samples,), optional
             Weights applied to individual samples.
@@ -2302,7 +2304,8 @@ def partial_fit(self, X, y=None, sample_weight=None):
 
         Returns
         -------
-        self : returns an instance of self.
+        self : object
+            Returns a fitted instance of self.
         """
 
         alpha = self.nu / 2
@@ -2383,6 +2386,8 @@ def fit(self, X, y=None, coef_init=None, offset_init=None, sample_weight=None):
         ----------
         X : {array-like, sparse matrix}, shape (n_samples, n_features)
             Training data.
+        y : Ignored
+            Not used, present for API consistency by convention.
 
         coef_init : array, shape (n_classes, n_features)
             The initial coefficients to warm-start the optimization.
@@ -2398,7 +2403,8 @@ def fit(self, X, y=None, coef_init=None, offset_init=None, sample_weight=None):
 
         Returns
         -------
-        self : returns an instance of self.
+        self : object
+            Returns a fitted instance of self.
         """
 
         alpha = self.nu / 2