ENH add possibility to have a callable for verbose_feature_names_out of ColumnTransformer

sklearn/compose/_column_transformer.py

@@ -10,6 +10,7 @@
 import warnings
 from collections import Counter, UserList
 from itertools import chain
+from functools import partial
 from numbers import Integral, Real
 
 import numpy as np
@@ -133,16 +134,29 @@ class ColumnTransformer(TransformerMixin, _BaseComposition):
  If True, the time elapsed while fitting each transformer will be
  printed as it is completed.
 
- verbose_feature_names_out : bool, default=True
+ verbose_feature_names_out : bool | str | Callable[[str, str], str], default=True
  If True, :meth:`ColumnTransformer.get_feature_names_out` will prefix
  all feature names with the name of the transformer that generated that
- feature.
+ feature. It is equivalent to setting
+ `verbose_feature_names_out="{transformer_name}__{feature_name}"`.
  If False, :meth:`ColumnTransformer.get_feature_names_out` will not
  prefix any feature names and will error if feature names are not
  unique.
+ If Callable[[str, str], str], :meth:`ColumnTransformer.get_feature_names_out`
+ will rename all the features using the name of the transformer. The
+ first argument of the callable is the transformer name and the
+ second argument is the feature name. The returned string will be the
+ new feature name.
+ If str, it must be a string ready for formatting. The given string will
+ be formatted using two field names: transformer_name and feature_name.
+ See str.format method from the standard library for more info.
 
  .. versionadded:: 1.0
 
+ .. versionchanged:: 1.X
+ `verbose_feature_names_out` can be a callable or a string to be formatted.
+
+
  force_int_remainder_cols : bool, default=True
  Force the columns of the last entry of `transformers_`, which
  corresponds to the "remainder" transformer, to always be stored as
@@ -283,7 +297,7 @@ class ColumnTransformer(TransformerMixin, _BaseComposition):
  "n_jobs": [Integral, None],
  "transformer_weights": [dict, None],
  "verbose": ["verbose"],
- "verbose_feature_names_out": ["boolean"],
+ "verbose_feature_names_out": ["boolean", str, callable],
  "force_int_remainder_cols": ["boolean"],
  }
 
@@ -655,11 +669,21 @@ def _add_prefix_for_feature_names_out(self, transformer_with_feature_names_out):
  feature_names_out : ndarray of shape (n_features,), dtype=str
  Transformed feature names.
  """
- if self.verbose_feature_names_out:
+ feature_names_out_callable = None
+ if callable(self.verbose_feature_names_out):
+ feature_names_out_callable = self.verbose_feature_names_out
+ elif isinstance(self.verbose_feature_names_out, str):
+ feature_names_out_callable = partial(
+ _feature_names_out, str_format=self.verbose_feature_names_out
+ )
+ elif self.verbose_feature_names_out is True:
+ feature_names_out_callable = _feature_names_out
+
+ if feature_names_out_callable is not None:
  # Prefix the feature names out with the transformers name
  names = list(
  chain.from_iterable(
- (f"{name}__{i}" for i in feature_names_out)
+ (feature_names_out_callable(name, i) for i in feature_names_out)
  for name, feature_names_out in transformer_with_feature_names_out
  )
  )
@@ -1652,3 +1676,13 @@ def _with_dtype_warning_enabled_set_to(warning_enabled, transformers):
  )
  result.append((name, trans, columns))
  return result
+
+
+def _feature_names_out(
+ transformer_name: str, feature_name: str, str_format: str | None = None
+) -> str:
+ if str_format is None:
+ str_format = "{transformer_name}__{feature_name}"
+ return str_format.format(
+ transformer_name=transformer_name, feature_name=feature_name
+ )

sklearn/compose/tests/test_column_transformer.py

@@ -1865,6 +1865,72 @@ def test_verbose_feature_names_out_true(transformers, remainder, expected_names)
  assert_array_equal(names, expected_names)
 
 
+def _feature_names_out_callable_name_clash(trans_name: str, feat_name: str):
+ return f"{trans_name[:2]}++{feat_name}"
+
+
+def _feature_names_out_callable_upper(trans_name: str, feat_name: str):
+ return f"{trans_name.upper()}={feat_name.upper()}"
+
+
+@pytest.mark.parametrize(
+ "transformers, remainder, verbose_feature_names_out, expected_names",
+ [
+ (
+ [
+ ("bycol1", TransWithNames(), ["d", "c"]),
+ ("bycol2", "passthrough", ["d"]),
+ ],
+ "passthrough",
+ _feature_names_out_callable_name_clash,
+ ["by++d", "by++c", "by++d", "re++a", "re++b"],
+ ),
+ (
+ [
+ ("bycol1", TransWithNames(), ["d", "c"]),
+ ("bycol2", "passthrough", ["d"]),
+ ],
+ "drop",
+ "{feature_name}-{transformer_name}",
+ ["d-bycol1", "c-bycol1", "d-bycol2"],
+ ),
+ (
+ [
+ ("bycol1", TransWithNames(), ["d", "c"]),
+ ("bycol2", "passthrough", slice("c", "d")),
+ ],
+ "passthrough",
+ _feature_names_out_callable_upper,
+ [
+ "BYCOL1=D",
+ "BYCOL1=C",
+ "BYCOL2=C",
+ "BYCOL2=D",
+ "REMAINDER=A",
+ "REMAINDER=B",
+ ],
+ ),
+ ],
+)
+def test_verbose_feature_names_out_callable_or_str(
+ transformers, remainder, verbose_feature_names_out, expected_names
+):
+ """Check feature_names_out for verbose_feature_names_out=True (default)"""
+ pd = pytest.importorskip("pandas")
+ df = pd.DataFrame([[1, 2, 3, 4]], columns=["a", "b", "c", "d"])
+ ct = ColumnTransformer(
+ transformers,
+ remainder=remainder,
+ verbose_feature_names_out=verbose_feature_names_out,
+ )
+ ct.fit(df)
+
+ names = ct.get_feature_names_out()
+ assert isinstance(names, np.ndarray)
+ assert names.dtype == object
+ assert_array_equal(names, expected_names)
+
+
 @pytest.mark.parametrize(
  "transformers, remainder, expected_names",
  [