FIX: avoid densifying sparse matrix before inverse_transform of OHE (#495)

Author: glemaitre

doc/whats_new/v0.4.rst

@@ -18,6 +18,10 @@ Bug fixes
   target or multilabel targets. Imbalanced-learn does not support this case.
   By :user:`Guillaume Lemaitre <glemaitre>` in :issue:`490`.
 
+- Fix a bug in :class:`imblearn.over_sampling.SMOTENC` in which an sparse
+  matrices were densify during ``inverse_transform``.
+  By :user:`Guillaume Lemaitre <glemaitre>` in :issue:`495`.
+
 Version 0.4
 ===========
 
@@ -219,4 +223,4 @@ Deprecation
 - Deprecate ``return_indices`` in all samplers. Instead, an attribute
   ``sample_indices_`` is created whenever the sampler is selecting a subset of
   the original samples.
-  :issue:`474` by :user:`Guillaume Lemaitre <glemaitre`.
+  :issue:`474` by :user:`Guillaume Lemaitre <glemaitre`.

imblearn/over_sampling/_smote.py

@@ -983,7 +983,7 @@ def _fit_resample(self, X, y):
         # reverse the encoding of the categorical features
         X_res_cat = X_resampled[:, self.continuous_features_.size:]
         X_res_cat.data = np.ones_like(X_res_cat.data)
-        X_res_cat_dec = self.ohe_.inverse_transform(X_res_cat.toarray())
+        X_res_cat_dec = self.ohe_.inverse_transform(X_res_cat)
 
         if sparse.issparse(X):
             X_resampled = sparse.hstack(