TST Extend tests for scipy.sparse/*array in sklearn/impute/tests/test_common (scikit-learn#27277)

Co-authored-by: Guillaume Lemaitre <g.lemaitre58@gmail.com>

Author: Charlie-XIAO

doc/whats_new/v1.4.rst

@@ -177,6 +177,14 @@ Changelog
 - |Fix| :func:`feature_selection.mutual_info_regression` now correctly computes the
   result when `X` is of integer dtype. :pr:`26748` by :user:`Yao Xiao <Charlie-XIAO>`.
 
+:mod:`sklearn.impute`
+.....................
+
+- |Enhancement| In :class:`impute.SimpleImputer`, :class:`impute.IterativeImputer`, and
+  :class:`impute.KNNImputer` with ``add_indicator=True``, using sparse arrays now
+  behaves in consistent with using sparse matrices in the `transform` and
+  `fit_transform` methods. :pr:`27277` by :user:`Yao Xiao <Charlie-XIAO>`.
+
 :mod:`sklearn.kernel_approximation`
 ...................................
 

sklearn/impute/_base.py

@@ -5,6 +5,7 @@
 import numbers
 import warnings
 from collections import Counter
+from functools import partial
 
 import numpy as np
 import numpy.ma as ma
@@ -115,7 +116,13 @@ def _concatenate_indicator(self, X_imputed, X_indicator):
         if not self.add_indicator:
             return X_imputed
 
-        hstack = sp.hstack if sp.issparse(X_imputed) else np.hstack
+        if sp.issparse(X_imputed):
+            # sp.hstack may result in different formats between sparse arrays and
+            # matrices; specify the format to keep consistent behavior
+            hstack = partial(sp.hstack, format=X_imputed.format)
+        else:
+            hstack = np.hstack
+
         if X_indicator is None:
             raise ValueError(
                 "Data from the missing indicator are not provided. Call "

sklearn/impute/tests/test_common.py

@@ -1,6 +1,5 @@
 import numpy as np
 import pytest
-from scipy import sparse
 
 from sklearn.experimental import enable_iterative_imputer  # noqa
 from sklearn.impute import IterativeImputer, KNNImputer, SimpleImputer
@@ -9,6 +8,7 @@
     assert_allclose_dense_sparse,
     assert_array_equal,
 )
+from sklearn.utils.fixes import CSR_CONTAINERS
 
 
 def imputers():
@@ -69,16 +69,17 @@ def test_imputers_add_indicator(marker, imputer):
 @pytest.mark.parametrize(
     "imputer", sparse_imputers(), ids=lambda x: x.__class__.__name__
 )
-def test_imputers_add_indicator_sparse(imputer, marker):
-    X = sparse.csr_matrix(
+@pytest.mark.parametrize("csr_container", CSR_CONTAINERS)
+def test_imputers_add_indicator_sparse(imputer, marker, csr_container):
+    X = csr_container(
         [
             [marker, 1, 5, marker, 1],
             [2, marker, 1, marker, 2],
             [6, 3, marker, marker, 3],
             [1, 2, 9, marker, 4],
         ]
     )
-    X_true_indicator = sparse.csr_matrix(
+    X_true_indicator = csr_container(
         [
             [1.0, 0.0, 0.0, 1.0],
             [0.0, 1.0, 0.0, 1.0],