[MRG] [ENH] Add sample_indices_ for SMOTE/ADASYN classes

imblearn/over_sampling/_adasyn.py

@@ -166,6 +166,8 @@ def _fit_resample(self, X, y):
         X_resampled = [X.copy()]
         y_resampled = [y.copy()]
 
+        self._sample_indices = np.stack((np.arange(len(y)), np.zeros(len(y)))).T
+
         for class_sample, n_samples in self.sampling_strategy_.items():
             if n_samples == 0:
                 continue
@@ -207,6 +209,14 @@ def _fit_resample(self, X, y):
             diffs = X_class[nns[rows, cols]] - X_class[rows]
             steps = random_state.uniform(size=(n_samples, 1))
 
+            self._sample_indices = np.concatenate(
+                (
+                    np.stack((np.arange(len(y)), np.zeros(len(y)))).T,
+                    np.stack((rows, cols)).T,
+                ),
+                axis=0,
+            )
+
             if sparse.issparse(X):
                 sparse_func = type(X).__name__
                 steps = getattr(sparse, sparse_func)(steps)
@@ -231,3 +241,21 @@ def _more_tags(self):
         return {
             "X_types": ["2darray"],
         }
+
+    def get_sample_indices(self):
+        """Return a tuple of indexes of the samples used to generate the new point.
+
+            Usable with ADASYN.
+
+        Returns
+        -------
+        _sample_indices : ndarray of shape (mother_sample_index, random_neighbour_index)
+                If the sample belongs to original dataset:
+                    mother_sample : index of the original sample
+                    random_neighbour_index : index of the neighbour sample
+        """
+        try:
+            self._sample_indices
+        except AttributeError:
+            return None
+        return self._sample_indices

imblearn/over_sampling/_smote/base.py

@@ -160,6 +160,8 @@ def _generate_samples(self, X, nn_data, nn_num, rows, cols, steps):
         else:
             X_new = X[rows] + steps * diffs
 
+        self._sample_indices = np.stack((rows, nn_num[rows, cols])).T
+
         return X_new.astype(X.dtype)
 
     def _in_danger_noise(self, nn_estimator, samples, target_class, y, kind="danger"):
@@ -365,8 +367,37 @@ def _fit_resample(self, X, y):
             X_resampled = np.vstack(X_resampled)
         y_resampled = np.hstack(y_resampled)
 
+        try:
+            self._sample_indices
+        except AttributeError:
+            self._sample_indices = np.stack((np.arange(len(y)), np.zeros(len(y)))).T
+            return X_resampled, y_resampled
+
+        self._sample_indices = np.concatenate(
+            (np.stack((np.arange(len(y)), np.zeros(len(y)))).T, self._sample_indices),
+            axis=0,
+        )
+
         return X_resampled, y_resampled
 
+    def get_sample_indices(self):
+        """Return a tuple of indexes of the samples used to generate the new point.
+
+            Usable with SMOTE.
+
+        Returns
+        -------
+        _sample_indices : ndarray of shape (mother_sample_index, random_neighbour_index)
+            If the sample belongs to original dataset:
+                mother_sample : index of the original sample
+                random_neighbour_index : index of the neighbour sample
+        """
+        try:
+            self._sample_indices
+        except AttributeError:
+            return None
+        return self._sample_indices
+
 
 @Substitution(
     sampling_strategy=BaseOverSampler._sampling_strategy_docstring,

imblearn/over_sampling/_smote/tests/test_borderline_smote.py

@@ -51,3 +51,24 @@ def test_borderline_smote(kind, data):
 
     assert_allclose(X_res_1, X_res_2)
     assert_array_equal(y_res_1, y_res_2)
+
+
+@pytest.mark.parametrize("kind", ["borderline-1", "borderline-2"])
+def test_borderline_smote_FutureWarning(kind, data):
+    bsmote = BorderlineSMOTE(kind=kind, random_state=42, n_jobs=1)
+    bsmote_nn = BorderlineSMOTE(
+        kind=kind,
+        random_state=42,
+        k_neighbors=NearestNeighbors(n_neighbors=6),
+        m_neighbors=NearestNeighbors(n_neighbors=11),
+    )
+    with pytest.warns(FutureWarning) as record:
+        bsmote.fit_resample(*data)
+        bsmote_nn.fit_resample(*data)
+    assert len(record) == 1
+    assert (
+        record[0].message.args[0]
+        == "The parameter `n_jobs` has been deprecated in 0.10"
+        " and will be removed in 0.12. You can pass an nearest"
+        " neighbors estimator where `n_jobs` is already set instead."
+    )

imblearn/over_sampling/_smote/tests/test_smote.py

@@ -37,6 +37,24 @@
 Y = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0])
 R_TOL = 1e-4
 
+XX = np.array(
+    [
+        [0.11622591, -0.0317206],
+        [0.77481731, 0.60935141],
+        [1.25192108, -0.22367336],
+        [0.53366841, -0.30312976],
+        [1.52091956, -0.49283504],
+        [-0.28162401, -2.10400981],
+        [0.83680821, 1.72827342],
+        [0.3084254, 0.33299982],
+        [0.70472253, -0.73309052],
+        [0.28893132, -0.38761769],
+        [1.15514042, 0.0129463],
+        [0.88407872, 0.35454207],
+    ]
+)
+YY = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 0, 0, 1])
+
 
 def test_sample_regular():
     smote = SMOTE(random_state=RND_SEED)
@@ -147,3 +165,67 @@ def test_sample_regular_with_nn():
     )
     assert_allclose(X_resampled, X_gt, rtol=R_TOL)
     assert_array_equal(y_resampled, y_gt)
+
+
+def test_sample_indices():
+    smote = SMOTE(random_state=RND_SEED)
+    smote.fit_resample(X, Y)
+    indices = smote.get_sample_indices()
+    indices_gt = np.array(
+        [
+            [0, 0],
+            [1, 0],
+            [2, 0],
+            [3, 0],
+            [4, 0],
+            [5, 0],
+            [6, 0],
+            [7, 0],
+            [8, 0],
+            [9, 0],
+            [10, 0],
+            [11, 0],
+            [12, 0],
+            [13, 0],
+            [14, 0],
+            [15, 0],
+            [16, 0],
+            [17, 0],
+            [18, 0],
+            [19, 0],
+            [0, 2],
+            [0, 1],
+            [0, 1],
+            [7, 2],
+        ]
+    )
+    assert_array_equal(indices, indices_gt)
+
+
+def test_sample_indices_balanced_dataset():
+    smote = SMOTE(random_state=RND_SEED)
+    smote.fit_resample(XX, YY)
+    indices = smote.get_sample_indices()
+    indices_gt = np.array(
+        [
+            [0, 0],
+            [1, 0],
+            [2, 0],
+            [3, 0],
+            [4, 0],
+            [5, 0],
+            [6, 0],
+            [7, 0],
+            [8, 0],
+            [9, 0],
+            [10, 0],
+            [11, 0],
+        ]
+    )
+    assert_array_equal(indices, indices_gt)
+
+
+def test_sample_indices_is_none():
+    smote = SMOTE(random_state=RND_SEED)
+    indices = smote.get_sample_indices()
+    assert_array_equal(indices, None)

imblearn/over_sampling/_smote/tests/test_smoten.py

@@ -52,3 +52,40 @@ def test_smoten_resampling():
     X_generated, y_generated = X_res[X.shape[0] :], y_res[X.shape[0] :]
     np.testing.assert_array_equal(X_generated, "blue")
     np.testing.assert_array_equal(y_generated, "not apple")
+
+
+def test_smoten_FutureWarning(data):
+    # check that SMOTEN throws FutureWarning for "n_jobs" and "keepdims"
+    X, y = data
+    sampler = SMOTEN(random_state=0, n_jobs=0)
+    with pytest.warns(FutureWarning) as record:
+        sampler.fit_resample(X, y)
+    assert (
+        record[0].message.args[0]
+        == "The parameter `n_jobs` has been deprecated in 0.10"
+        " and will be removed in 0.12. You can pass an nearest"
+        " neighbors estimator where `n_jobs` is already set instead."
+    )
+
+
+@pytest.fixture
+def data_balanced():
+    rng = np.random.RandomState(0)
+
+    feature_1 = ["A"] * 10 + ["B"] * 20 + ["C"] * 30
+    feature_2 = ["A"] * 40 + ["B"] * 20
+    feature_3 = ["A"] * 20 + ["B"] * 20 + ["C"] * 10 + ["D"] * 10
+    X = np.array([feature_1, feature_2, feature_3], dtype=object).T
+    rng.shuffle(X)
+    y = np.array([0] * 30 + [1] * 30, dtype=np.int32)
+    y_labels = np.array(["not apple", "apple"], dtype=object)
+    y = y_labels[y]
+    return X, y
+
+
+def test_smoten_balanced_data(data_balanced):
+    X, y = data_balanced
+    sampler = SMOTEN(random_state=0)
+    X_res, y_res = sampler.fit_resample(X, y)
+    assert X_res.shape == (60, 3)
+    assert y_res.shape == (60,)

imblearn/over_sampling/tests/test_adasyn.py

@@ -37,6 +37,37 @@
 Y = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0])
 R_TOL = 1e-4
 
+XX = np.array(
+    [
+        [0.11622591, -0.0317206],
+        [0.77481731, 0.60935141],
+        [1.25192108, -0.22367336],
+        [0.53366841, -0.30312976],
+        [1.52091956, -0.49283504],
+        [-0.28162401, -2.10400981],
+        [0.83680821, 1.72827342],
+        [0.3084254, 0.33299982],
+        [0.70472253, -0.73309052],
+        [0.28893132, -0.38761769],
+        [1.15514042, 0.0129463],
+        [0.88407872, 0.35454207],
+    ]
+)
+YY = np.array([0, 1, 0, 0, 0, 1, 0, 1, 1, 1, 1, 0])
+
+XXX = np.array(
+    [
+        [0.915, 0.892],
+        [0.926, 0.959],
+        [0.917, 0.983],
+        [0.945, 0.967],
+        [-0.844, -0.925],
+        [-0.987, -0.946],
+        [-0.962, -0.948],
+    ]
+)
+YYY = np.array([1, 1, 1, 1, 0, 0, 0])
+
 
 def test_ada_init():
     sampling_strategy = "auto"