scikit-learn · rth · Feb 12, 2021 · Nov 7, 2020 · Feb 12, 2021 · Feb 12, 2021
diff --git a/sklearn/covariance/_elliptic_envelope.py b/sklearn/covariance/_elliptic_envelope.py
@@ -37,7 +37,7 @@ class EllipticEnvelope(OutlierMixin, MinCovDet):
 
     contamination : float, default=0.1
         The amount of contamination of the data set, i.e. the proportion
-        of outliers in the data set. Range is (0, 0.5).
+        of outliers in the data set. Range is (0, 0.5].
 
     random_state : int, RandomState instance or None, default=None
         Determines the pseudo random number generator for shuffling
@@ -142,6 +142,11 @@ def fit(self, X, y=None):
         y : Ignored
             Not used, present for API consistency by convention.
         """
+        if self.contamination != 'auto':
+            if not(0. < self.contamination <= .5):
+                raise ValueError("contamination must be in (0, 0.5], "
+                                 "got: %f" % self.contamination)
+
         super().fit(X)
         self.offset_ = np.percentile(-self.dist_, 100. * self.contamination)
         return self

diff --git a/sklearn/ensemble/_iforest.py b/sklearn/ensemble/_iforest.py
@@ -70,7 +70,7 @@ class IsolationForest(OutlierMixin, BaseBagging):
 
             - If 'auto', the threshold is determined as in the
               original paper.
-            - If float, the contamination should be in the range [0, 0.5].
+            - If float, the contamination should be in the range (0, 0.5].
 
         .. versionchanged:: 0.22
            The default value of ``contamination`` changed from 0.1
@@ -250,6 +250,11 @@ def fit(self, X, y=None, sample_weight=None):
         # ensure that max_sample is in [1, n_samples]:
         n_samples = X.shape[0]
 
+        if self.contamination != 'auto':
+            if not(0. < self.contamination <= .5):
+                raise ValueError("contamination must be in (0, 0.5], "
+                                 "got: %f" % self.contamination)
+
         if isinstance(self.max_samples, str):
             if self.max_samples == 'auto':
                 max_samples = min(256, n_samples)

diff --git a/sklearn/neighbors/_lof.py b/sklearn/neighbors/_lof.py
@@ -104,7 +104,7 @@ class LocalOutlierFactor(KNeighborsMixin,
 
         - if 'auto', the threshold is determined as in the
           original paper,
-        - if a float, the contamination should be in the range [0, 0.5].
+        - if a float, the contamination should be in the range (0, 0.5].
 
         .. versionchanged:: 0.22
            The default value of ``contamination`` changed from 0.1

diff --git a/sklearn/utils/estimator_checks.py b/sklearn/utils/estimator_checks.py
@@ -2071,9 +2071,10 @@ def check_outliers_train(name, estimator_orig, readonly_memmap=True):
             check_outlier_corruption(num_outliers, expected_outliers, decision)
 
         # raises error when contamination is a scalar and not in [0,1]
+        msg = r"contamination must be in \(0, 0.5]"
         for contamination in [-0.5, 2.3]:
             estimator.set_params(contamination=contamination)
-            with raises(ValueError):
+            with raises(ValueError, match=msg):
                 estimator.fit(X)
 
 
@@ -2964,9 +2965,10 @@ def check_outliers_fit_predict(name, estimator_orig):
             check_outlier_corruption(num_outliers, expected_outliers, decision)
 
         # raises error when contamination is a scalar and not in [0,1]
-        for contamination in [-0.5, 2.3]:
+        msg = r"contamination must be in \(0, 0.5]"
+        for contamination in [-0.5, -0.001, 0.5001, 2.3]:
             estimator.set_params(contamination=contamination)
-            with raises(ValueError):
+            with raises(ValueError, match=msg):
                 estimator.fit_predict(X)