unit tests added for safe-level SMOTE

Author: laurallu

imblearn/over_sampling/_smote.py

@@ -1324,18 +1324,18 @@ class SLSMOTE(BaseSMOTE):
 
     Parameters
     -----------
-     {sampling_strategy}
+    {sampling_strategy}
 
-     {random_state}
+    {random_state}
 
-     k_neighbors : int or object, optional (default=5)
+    k_neighbors : int or object, optional (default=5)
         If ``int``, number of nearest neighbours to used to construct synthetic
         samples.  If object, an estimator that inherits from
         :class:`sklearn.neighbors.base.KNeighborsMixin` that will be used to
         find the k_neighbors.
 
     m_neighbors : int or object, optional (default=10)
-        If ``int``, number of nearest neighbours to use to determine the safe
+        If ``int``, number of nearest neighbours used to determine the safe
         level of an instance. If object, an estimator that inherits from
         :class:`sklearn.neighbors.base.KNeighborsMixin` that will be used
         to find the m_neighbors.
@@ -1582,16 +1582,16 @@ def _generate_gap(self, a_ratio, rand_state=None):
 
         Parameters
         ----------
-         a_ratio: float
-                  sl_ratio of a single data point
+        a_ratio: float
+                 sl_ratio of a single data point
 
-         rand_state: random state object or int
+        rand_state: random state object or int
 
 
-         Returns
-         ------------
-         gap: float
-              a number between 0 and 1
+        Returns
+        ------------
+        gap: float
+             a number between 0 and 1
 
         """
 

imblearn/over_sampling/tests/test_sl_smote.py

@@ -0,0 +1,66 @@
+import pytest
+import numpy as np
+
+from sklearn.neighbors import NearestNeighbors
+from scipy import sparse
+
+from sklearn.utils._testing import assert_allclose
+from sklearn.utils._testing import assert_array_equal
+
+from imblearn.over_sampling import SLSMOTE
+
+
+def data_np():
+    rng = np.random.RandomState(42)
+    X = rng.randn(20, 2)
+    y = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0])
+    return X, y
+
+
+def data_sparse(format):
+    X = sparse.random(20, 2, density=0.3, format=format, random_state=42)
+    y = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0])
+    return X, y
+
+
+@pytest.mark.parametrize(
+    "data",
+    [data_np(), data_sparse('csr'), data_sparse('csc')]
+)
+def test_slsmote(data):
+    y_gt = np.array([0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0,
+                     0, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0])
+    X, y = data
+    slsmote = SLSMOTE(random_state=42)
+    X_res, y_res = slsmote.fit_resample(X, y)
+
+    assert X_res.shape == (24, 2)
+    assert_array_equal(y_res, y_gt)
+
+
+def test_slsmote_nn():
+    X, y = data_np()
+    slsmote = SLSMOTE(random_state=42)
+    slsmote_nn = SLSMOTE(
+        random_state=42,
+        k_neighbors=NearestNeighbors(n_neighbors=6),
+        m_neighbors=NearestNeighbors(n_neighbors=11),
+    )
+
+    X_res_1, y_res_1 = slsmote.fit_resample(X, y)
+    X_res_2, y_res_2 = slsmote_nn.fit_resample(X, y)
+
+    assert_allclose(X_res_1, X_res_2)
+    assert_array_equal(y_res_1, y_res_2)
+
+
+def test_slsmote_pd():
+    pd = pytest.importorskip("pandas")
+    X, y = data_np()
+    X_pd = pd.DataFrame(X)
+    slsmote = SLSMOTE(random_state=42)
+    X_res, y_res = slsmote.fit_resample(X, y)
+    X_res_pd, y_res_pd = slsmote.fit_resample(X_pd, y)
+
+    assert X_res_pd.tolist() == X_res.tolist()
+    assert_allclose(y_res_pd, y_res)