MNT improve error message in _num_samples (scikit-learn#30355)

Co-authored-by: Loïc Estève <[email protected]>

Author: StefanieSenger

sklearn/tree/tests/test_tree.py

@@ -6,6 +6,7 @@
 import copyreg
 import io
 import pickle
+import re
 import struct
 from itertools import chain, product
 
@@ -1137,7 +1138,13 @@ def test_sample_weight_invalid():
         clf.fit(X, y, sample_weight=sample_weight)
 
     sample_weight = np.array(0)
-    expected_err = r"Singleton.* cannot be considered a valid collection"
+
+    expected_err = re.escape(
+        (
+            "Input should have at least 1 dimension i.e. satisfy "
+            "`len(x.shape) > 0`, got scalar `array(0.)` instead."
+        )
+    )
     with pytest.raises(TypeError, match=expected_err):
         clf.fit(X, y, sample_weight=sample_weight)
 

sklearn/utils/tests/test_validation.py

@@ -743,7 +743,12 @@ def test_check_array_min_samples_and_features_messages():
         check_array([], ensure_2d=False)
 
     # Invalid edge case when checking the default minimum sample of a scalar
-    msg = r"Singleton array array\(42\) cannot be considered a valid" " collection."
+    msg = re.escape(
+        (
+            "Input should have at least 1 dimension i.e. satisfy "
+            "`len(x.shape) > 0`, got scalar `array(42)` instead."
+        )
+    )
     with pytest.raises(TypeError, match=msg):
         check_array(42, ensure_2d=False)
 

sklearn/utils/validation.py

@@ -397,7 +397,8 @@ def _num_samples(x):
     if hasattr(x, "shape") and x.shape is not None:
         if len(x.shape) == 0:
             raise TypeError(
-                "Singleton array %r cannot be considered a valid collection." % x
+                "Input should have at least 1 dimension i.e. satisfy "
+                f"`len(x.shape) > 0`, got scalar `{x!r}` instead."
             )
         # Check that shape is returning an integer or default to len
         # Dask dataframes may not return numeric shape[0] value