Enforce to get same data at fitting and sampling

Author: Guillaume Lemaitre

unbalanced_dataset/base.py

@@ -73,6 +73,7 @@ def __init__(self, ratio='auto', random_state=None, verbose=True):
         self.min_c_ = None
         self.maj_c_ = None
         self.stats_c_ = {}
+        self.X_shape_ = None
 
     @abstractmethod
     def fit(self, X, y):
@@ -110,6 +111,10 @@ def fit(self, X, y):
             warnings.warn('Only one class detected, something will get wrong',
                           RuntimeWarning)
 
+        # Store the size of X to check at sampling time if we have the
+        # same data
+        self.X_shape_ = X.shape
+
         # Create a dictionary containing the class statistics
         self.stats_c_ = Counter(y)
 
@@ -157,6 +162,12 @@ def sample(self, X, y):
         if not self.stats_c_:
             raise RuntimeError('You need to fit the data, first!!!')
 
+        # Check if the size of the data is identical than at fitting
+        if X.shape != self.X_shape_:
+            raise RuntimeError('The data that you attempt to resample do not'
+                               ' seem to be the one earlier fitted. Use the'
+                               ' fitted data.')
+
         return self
 
     def fit_sample(self, X, y):

unbalanced_dataset/combine/smote_enn.py

@@ -79,6 +79,9 @@ class SMOTEENN(SamplerMixin):
         A dictionary in which the number of occurences of each class is
         reported.
 
+    X_shape_ : tuple of int
+        Shape of the data `X` during fitting.
+
     Notes
     -----
     The method is presented in [1]_.

unbalanced_dataset/combine/smote_tomek.py

@@ -80,6 +80,9 @@ class SMOTETomek(SamplerMixin):
         A dictionary in which the number of occurences of each class is
         reported.
 
+    X_shape_ : tuple of int
+        Shape of the data `X` during fitting.
+
     Notes
     -----
     The methos is presented in [1]_.

unbalanced_dataset/combine/tests/test_smote_enn.py

@@ -116,3 +116,14 @@ def test_sample_regular_half():
     y_gt = np.load(os.path.join(currdir, 'data', 'smote_enn_reg_y_05.npy'))
     assert_array_equal(X_resampled, X_gt)
     assert_array_equal(y_resampled, y_gt)
+
+
+def test_sample_wrong_X():
+    """Test either if an error is raised when X is different at fitting
+    and sampling"""
+
+    # Create the object
+    sm = SMOTEENN(random_state=RND_SEED)
+    sm.fit(X, Y)
+    assert_raises(RuntimeError, sm.sample, np.random.random((100, 40)),
+                  np.array([0] * 50 + [1] * 50))

unbalanced_dataset/combine/tests/test_smote_tomek.py

@@ -116,3 +116,14 @@ def test_sample_regular_half():
     y_gt = np.load(os.path.join(currdir, 'data', 'smote_tomek_reg_y_05.npy'))
     assert_array_equal(X_resampled, X_gt)
     assert_array_equal(y_resampled, y_gt)
+
+
+def test_sample_wrong_X():
+    """Test either if an error is raised when X is different at fitting
+    and sampling"""
+
+    # Create the object
+    sm = SMOTETomek(random_state=RND_SEED)
+    sm.fit(X, Y)
+    assert_raises(RuntimeError, sm.sample, np.random.random((100, 40)),
+                  np.array([0] * 50 + [1] * 50))

unbalanced_dataset/ensemble/balance_cascade.py

@@ -71,6 +71,9 @@ class BalanceCascade(EnsembleSampler):
         A dictionary in which the number of occurences of each class is
         reported.
 
+    X_shape_ : tuple of int
+        Shape of the data `X` during fitting.
+
     Notes
     -----
     The method is described in [1]_.

unbalanced_dataset/ensemble/easy_ensemble.py

@@ -60,6 +60,9 @@ class EasyEnsemble(EnsembleSampler):
         A dictionary in which the number of occurences of each class is
         reported.
 
+    X_shape_ : tuple of int
+        Shape of the data `X` during fitting.
+
     Notes
     -----
     The method is described in [1]_.

unbalanced_dataset/ensemble/tests/test_balance_cascade.py

@@ -331,3 +331,14 @@ def test_fit_sample_auto_early_stop():
         assert_array_equal(X_resampled[idx], X_gt[idx])
         assert_array_equal(y_resampled[idx], y_gt[idx])
         assert_array_equal(idx_under[idx], idx_gt[idx])
+
+
+def test_sample_wrong_X():
+    """Test either if an error is raised when X is different at fitting
+    and sampling"""
+
+    # Create the object
+    bc = BalanceCascade(random_state=RND_SEED)
+    bc.fit(X, Y)
+    assert_raises(RuntimeError, bc.sample, np.random.random((100, 40)),
+                  np.array([0] * 50 + [1] * 50))

unbalanced_dataset/ensemble/tests/test_easy_ensemble.py

@@ -160,3 +160,14 @@ def test_fit_sample_half():
     y_gt = np.load(os.path.join(currdir, 'data', 'ee_y_05.npy'))
     assert_array_equal(X_resampled, X_gt)
     assert_array_equal(y_resampled, y_gt)
+
+
+def test_sample_wrong_X():
+    """Test either if an error is raised when X is different at fitting
+    and sampling"""
+
+    # Create the object
+    ee = EasyEnsemble(random_state=RND_SEED)
+    ee.fit(X, Y)
+    assert_raises(RuntimeError, ee.sample, np.random.random((100, 40)),
+                  np.array([0] * 50 + [1] * 50))

unbalanced_dataset/over_sampling/random_over_sampler.py

@@ -52,6 +52,9 @@ class RandomOverSampler(OverSampler):
         A dictionary in which the number of occurences of each class is
         reported.
 
+    X_shape_ : tuple of int
+        Shape of the data `X` during fitting.
+
     Notes
     -----
     Supports multiple classes.