ENH support array-like of str for categorical_features in SMOTENC (#1008)

Author: glemaitre

doc/over_sampling.rst

@@ -192,8 +192,8 @@ which categorical data are treated differently::
 
 In this data set, the first and last features are considered as categorical
 features. One needs to provide this information to :class:`SMOTENC` via the
-parameters ``categorical_features`` either by passing the indices of these
-features or a boolean mask marking these features::
+parameters ``categorical_features`` either by passing the indices, the feature
+names when `X` is a pandas DataFrame, or a boolean mask marking these features::
 
   >>> from imblearn.over_sampling import SMOTENC
   >>> smote_nc = SMOTENC(categorical_features=[0, 2], random_state=0)

doc/whats_new/v0.11.rst

@@ -57,3 +57,7 @@ Enhancements
   :class:`~imblearn.over_sampling.RandomOverSampler` (when `shrinkage is not
   None`) now accept any data types and will not attempt any data conversion.
   :pr:`1004` by :user:`Guillaume Lemaitre <glemaitre>`.
+
+- :class:`~imblearn.over_sampling.SMOTENC` now support passing array-like of `str`
+  when passing the `categorical_features` parameter.
+  :pr:`1008` by :user`Guillaume Lemaitre <glemaitre>`.

imblearn/over_sampling/_smote/base.py

@@ -16,7 +16,12 @@
 from sklearn.base import clone
 from sklearn.exceptions import DataConversionWarning
 from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder
-from sklearn.utils import _safe_indexing, check_array, check_random_state
+from sklearn.utils import (
+    _get_column_indices,
+    _safe_indexing,
+    check_array,
+    check_random_state,
+)
 from sklearn.utils.sparsefuncs_fast import (
     csc_mean_variance_axis0,
     csr_mean_variance_axis0,
@@ -390,10 +395,14 @@ class SMOTENC(SMOTE):
 
     Parameters
     ----------
-    categorical_features : array-like of shape (n_cat_features,) or (n_features,)
+    categorical_features : array-like of shape (n_cat_features,) or (n_features,), \
+            dtype={{bool, int, str}}
         Specified which features are categorical. Can either be:
 
-        - array of indices specifying the categorical features;
+        - array of `int` corresponding to the indices specifying the categorical
+          features;
+        - array of `str` corresponding to the feature names. `X` should be a pandas
+          :class:`pandas.DataFrame` in this case.
         - mask array of shape (n_features, ) and ``bool`` dtype for which
           ``True`` indicates the categorical features.
 
@@ -565,24 +574,16 @@ def _check_X_y(self, X, y):
         self._check_feature_names(X, reset=True)
         return X, y, binarize_y
 
-    def _validate_estimator(self):
-        super()._validate_estimator()
-        categorical_features = np.asarray(self.categorical_features)
-        if categorical_features.dtype.name == "bool":
-            self.categorical_features_ = np.flatnonzero(categorical_features)
-        else:
-            if any(
-                [cat not in np.arange(self.n_features_) for cat in categorical_features]
-            ):
-                raise ValueError(
-                    f"Some of the categorical indices are out of range. Indices"
-                    f" should be between 0 and {self.n_features_ - 1}"
-                )
-            self.categorical_features_ = categorical_features
+    def _validate_column_types(self, X):
+        self.categorical_features_ = np.array(
+            _get_column_indices(X, self.categorical_features)
+        )
         self.continuous_features_ = np.setdiff1d(
             np.arange(self.n_features_), self.categorical_features_
         )
 
+    def _validate_estimator(self):
+        super()._validate_estimator()
         if self.categorical_features_.size == self.n_features_in_:
             raise ValueError(
                 "SMOTE-NC is not designed to work only with categorical "
@@ -600,6 +601,7 @@ def _fit_resample(self, X, y):
             )
 
         self.n_features_ = _num_features(X)
+        self._validate_column_types(X)
         self._validate_estimator()
 
         # compute the median of the standard deviation of the minority class

imblearn/over_sampling/_smote/tests/test_smote_nc.py

@@ -63,7 +63,7 @@ def data_heterogneous_masked():
     X[:, 3] = rng.randint(3, size=30)
     y = np.array([0] * 10 + [1] * 20)
     # return the categories
-    return X, y, [True, False, True]
+    return X, y, [True, False, False, True]
 
 
 def data_heterogneous_unordered_multiclass():
@@ -98,7 +98,7 @@ def test_smotenc_error():
     X, y, _ = data_heterogneous_unordered()
     categorical_features = [0, 10]
     smote = SMOTENC(random_state=0, categorical_features=categorical_features)
-    with pytest.raises(ValueError, match="indices are out of range"):
+    with pytest.raises(ValueError, match="all features must be in"):
         smote.fit_resample(X, y)
 
 
@@ -324,3 +324,28 @@ def test_smotenc_bool_categorical():
     X_res, y_res = smote.fit_resample(X, y)
     pd.testing.assert_series_equal(X_res.dtypes, X.dtypes)
     assert len(X_res) == len(y_res)
+
+
+def test_smotenc_categorical_features_str():
+    """Check that we support array-like of strings for `categorical_features` using
+    pandas dataframe.
+    """
+    pd = pytest.importorskip("pandas")
+
+    X = pd.DataFrame(
+        {
+            "A": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
+            "B": ["a", "b"] * 5,
+            "C": ["a", "b", "c"] * 3 + ["a"],
+        }
+    )
+    X = pd.concat([X] * 10, ignore_index=True)
+    y = np.array([0] * 70 + [1] * 30)
+    smote = SMOTENC(categorical_features=["B", "C"], random_state=0)
+    X_res, y_res = smote.fit_resample(X, y)
+    assert X_res["B"].isin(["a", "b"]).all()
+    assert X_res["C"].isin(["a", "b", "c"]).all()
+    counter = Counter(y_res)
+    assert counter[0] == counter[1] == 70
+    assert_array_equal(smote.categorical_features_, [1, 2])
+    assert_array_equal(smote.continuous_features_, [0])