some fixes

smote_variants/undersampling/_oversampling_driven_undersampling.py

@@ -59,25 +59,27 @@ def sample(self, X, y):
         n_maj, n_min = np.bincount(y)
 
         X_samp, _ = self.oversampler.sample(X, y)
-        X_new = X_samp[X.shape[0] :]
+        X_new = X_samp[X.shape[0]:]
         X_min = X[y == 1]
         X_maj = X[y == 0]
 
         dists = X_new - X_maj[:, None]
 
-        dists = np.mean(np.sqrt(np.sum((dists) ** 2, axis=2)), axis=1)
+        dists = np.min(np.sqrt(np.sum((dists) ** 2, axis=2)), axis=1)
 
         if self.mode == "random":
-            inv_dists = 1.0 / np.where(dists < 1e-8, 1e-8, dists)
-            p = dists / np.sum(inv_dists)
+            dists = (dists - np.min(dists)) / (np.max(dists) - np.min(dists))
+            inv_dists = 1.0 - dists
+
+            p = inv_dists / np.sum(inv_dists)
 
             mask = self.random_state.choice(np.arange(n_maj), n_min, p=p, replace=False)
             X_maj = X_maj[mask]
         elif self.mode == "farthest":
             sorting = np.argsort(dists)
-            X_maj = X_maj[sorting[:n_min]]
+            X_maj = X_maj[sorting][:n_min]
 
-        X_res = np.vstack([X_maj, X_min])  # pylint: disable=invalid-name
-        y_res = np.hstack([np.repeat(0, n_min), np.repeat(1, n_min)])
+        X_res = np.vstack([X_maj, X_min]).copy()  # pylint: disable=invalid-name
+        y_res = np.hstack([np.repeat(0, X_maj.shape[0]), np.repeat(1, X_min.shape[0])])
 
         return X_res, y_res

smote_variants/undersampling/_random_undersampling.py

@@ -42,7 +42,12 @@ def sample(self, X, y):
             np.array, np.array: the undersampled feature vectors and class labels
         """
         n_maj, n_min = np.bincount(y)
+        #mask = self.random_state.choice(np.arange(n_maj), n_min, p=np.repeat(1.0/n_maj, n_maj), replace=False)
         mask = self.random_state.choice(np.arange(n_maj), n_min, replace=False)
         X_maj = X[y == 0][mask]
+        X_min = X[y == 1]
 
-        return np.vstack([X_maj, X[y == 1]]), np.hstack([np.repeat(0, n_min), np.repeat(1, n_min)])
+        X_res = np.vstack([X_maj, X_min]).copy()  # pylint: disable=invalid-name
+        y_res = np.hstack([np.repeat(0, X_maj.shape[0]), np.repeat(1, X_min.shape[0])])
+
+        return X_res, y_res

tests/undersampling/test_undersampling.py

@@ -11,7 +11,6 @@
     RandomUndersampling,
     OversamplingDrivenUndersampling,
 )
-from smote_variants.oversampling import SMOTE
 
 undersamplings = [RandomUndersampling, OversamplingDrivenUndersampling]
 
@@ -37,3 +36,26 @@ def test_undersampling_normal(undersampling):
     _, y_samp = undersampling_obj.sample(X, y)
 
     assert np.sum(y_samp) == len(y_samp) - np.sum(y_samp)
+
+@pytest.mark.parametrize("undersampling", undersamplings)
+def test_undersampling_normal_farthest(undersampling):
+    """
+    Testing the undersampling farthest strategy
+
+    Args:
+        undersampling (cls): the undersampling class
+    """
+
+    dataset = load_normal()
+
+    X = dataset["data"]
+    y = dataset["target"]
+
+    params = undersampling.parameter_combinations()
+
+    for param in params:
+        undersampling_obj = undersampling(**param)
+
+        _, y_samp = undersampling_obj.sample(X, y)
+
+        assert np.sum(y_samp) == len(y_samp) - np.sum(y_samp)