ENH num_features for a 1d collection of dicts is undefined (#19740)

thomasjpfan · lorentzenchr · web-flow · commit 80f923e00d69 · 2021-03-29T16:12:54.000+02:00
Co-authored-by: Christian Lorentzen &lt;lorentzen.ch@gmail.com&gt;
diff --git a/sklearn/utils/tests/test_validation.py b/sklearn/utils/tests/test_validation.py
@@ -1345,9 +1345,10 @@ def test_num_features(constructor_name):
         [1, 2, 3],
         ["a", "b", "c"],
         [False, True, False],
-        [1.0, 3.4, 4.0]
+        [1.0, 3.4, 4.0],
+        [{"a": 1}, {"b": 2}, {"c": 3}],
     ],
-    ids=["int", "str", "bool", "float"]
+    ids=["int", "str", "bool", "float", "dict"]
 )
 @pytest.mark.parametrize("constructor_name", [
     "list", "tuple", "array", "series"
@@ -1368,6 +1369,8 @@ def test_num_features_errors_1d_containers(X, constructor_name):
         message += " with shape (3,)"
     elif isinstance(X[0], str):
         message += " where the samples are of type str"
+    elif isinstance(X[0], dict):
+        message += " where the samples are of type dict"
     with pytest.raises(TypeError, match=re.escape(message)):
         _num_features(X)
 
diff --git a/sklearn/utils/validation.py b/sklearn/utils/validation.py
@@ -226,8 +226,8 @@ def _num_features(X):
 
     first_sample = X[0]
 
-    # Do not consider an array-like of strings to be a 2D array
-    if isinstance(first_sample, (str, bytes)):
+    # Do not consider an array-like of strings or dicts to be a 2D array
+    if isinstance(first_sample, (str, bytes, dict)):
         message += (f" where the samples are of type "
                     f"{type(first_sample).__qualname__}")
         raise TypeError(message)
