ENH accept non finite values in random samplers (#643)

Author: glemaitre

doc/whats_new/v0.6.rst

@@ -51,6 +51,11 @@ Enhancement
   to check or not the input ``X`` and ``y``.
   :pr:`637` by :user:`Guillaume Lemaitre <glemaitre>`.
 
+- :class:`imblearn.under_sampling.RandomUnderSampler`,
+  :class:`imblearn.over_sampling.RandomOverSampler` can resample when non
+  finite values are present in ``X``.
+  :pr:`643` by `Guillaume Lemaitre <glemaitre>`.
+
 Deprecation
 ...........
 

imblearn/over_sampling/_random_over_sampler.py

@@ -79,7 +79,8 @@ def _check_X_y(X, y):
         y, binarize_y = check_target_type(y, indicate_one_vs_all=True)
         if not hasattr(X, "loc"):
             # Do not convert dataframe
-            X = check_array(X, accept_sparse=["csr", "csc"], dtype=None)
+            X = check_array(X, accept_sparse=["csr", "csc"], dtype=None,
+                            force_all_finite=False)
         y = check_array(
             y, accept_sparse=["csr", "csc"], dtype=None, ensure_2d=False
         )
@@ -108,4 +109,8 @@ def _fit_resample(self, X, y):
         )
 
     def _more_tags(self):
-        return {"X_types": ["2darray", "string"], "sample_indices": True}
+        return {
+            "X_types": ["2darray", "string"],
+            "sample_indices": True,
+            "allow_nan": True,
+        }

imblearn/over_sampling/tests/test_random_over_sampler.py

@@ -125,3 +125,23 @@ def test_random_over_sampling_heterogeneous_data():
     assert y_res.shape[0] == 4
     assert X_res.dtype == object
     assert X_res[-1, 0] in X_hetero[:, 0]
+
+
+def test_random_over_sampling_nan_inf():
+    # check that we can oversample even with missing or infinite data
+    # regression tests for #605
+    rng = np.random.RandomState(42)
+    n_not_finite = X.shape[0] // 3
+    row_indices = rng.choice(np.arange(X.shape[0]), size=n_not_finite)
+    col_indices = rng.randint(0, X.shape[1], size=n_not_finite)
+    not_finite_values = rng.choice([np.nan, np.inf], size=n_not_finite)
+
+    X_ = X.copy()
+    X_[row_indices, col_indices] = not_finite_values
+
+    ros = RandomOverSampler(random_state=0)
+    X_res, y_res = ros.fit_resample(X_, Y)
+
+    assert y_res.shape == (14,)
+    assert X_res.shape == (14, 2)
+    assert np.any(~np.isfinite(X_res))

imblearn/under_sampling/_prototype_selection/_random_under_sampler.py

@@ -85,7 +85,8 @@ def _check_X_y(X, y):
         y, binarize_y = check_target_type(y, indicate_one_vs_all=True)
         if not hasattr(X, "loc"):
             # Do not convert dataframe
-            X = check_array(X, accept_sparse=["csr", "csc"], dtype=None)
+            X = check_array(X, accept_sparse=["csr", "csc"], dtype=None,
+                            force_all_finite=False)
         y = check_array(
             y, accept_sparse=["csr", "csc"], dtype=None, ensure_2d=False
         )
@@ -121,4 +122,8 @@ def _fit_resample(self, X, y):
         return _safe_indexing(X, idx_under), _safe_indexing(y, idx_under)
 
     def _more_tags(self):
-        return {"X_types": ["2darray", "string"], "sample_indices": True}
+        return {
+            "X_types": ["2darray", "string"],
+            "sample_indices": True,
+            "allow_nan": True,
+        }

imblearn/under_sampling/_prototype_selection/tests/test_random_under_sampler.py

@@ -110,3 +110,23 @@ def test_random_under_sampling_heterogeneous_data():
     assert X_res.shape[0] == 2
     assert y_res.shape[0] == 2
     assert X_res.dtype == object
+
+
+def test_random_under_sampling_nan_inf():
+    # check that we can undersample even with missing or infinite data
+    # regression tests for #605
+    rng = np.random.RandomState(42)
+    n_not_finite = X.shape[0] // 3
+    row_indices = rng.choice(np.arange(X.shape[0]), size=n_not_finite)
+    col_indices = rng.randint(0, X.shape[1], size=n_not_finite)
+    not_finite_values = rng.choice([np.nan, np.inf], size=n_not_finite)
+
+    X_ = X.copy()
+    X_[row_indices, col_indices] = not_finite_values
+
+    rus = RandomUnderSampler(random_state=0)
+    X_res, y_res = rus.fit_resample(X_, Y)
+
+    assert y_res.shape == (6,)
+    assert X_res.shape == (6, 2)
+    assert np.any(~np.isfinite(X_res))