Backport PR pandas-dev#49864 on branch 1.5.x (REGR: DataFrameGroupBy.transform with reducer and as_index=False returns null values) (pandas-dev#49868)

* Backport PR pandas-dev#49864: REGR: DataFrameGroupBy.transform with reducer and as_index=False returns null values

* Add back request

Co-authored-by: Richard Shadrach <[email protected]>
Co-authored-by: Patrick Hoefler <[email protected]>

Author: 3 people

doc/source/whatsnew/v1.5.3.rst

@@ -14,6 +14,7 @@ including other versions of pandas.
 Fixed regressions
 ~~~~~~~~~~~~~~~~~
 - Fixed performance regression in :meth:`Series.isin` when ``values`` is empty (:issue:`49839`)
+- Fixed regression in :meth:`DataFrameGroupBy.transform` when used with ``as_index=False`` (:issue:`49834`)
 -
 
 .. ---------------------------------------------------------------------------

pandas/core/groupby/groupby.py

@@ -1838,7 +1838,10 @@ def _transform(self, func, *args, engine=None, engine_kwargs=None, **kwargs):
             # and deal with possible broadcasting below.
             # Temporarily set observed for dealing with categoricals.
             with com.temp_setattr(self, "observed", True):
-                result = getattr(self, func)(*args, **kwargs)
+                with com.temp_setattr(self, "as_index", True):
+                    # GH#49834 - result needs groups in the index for
+                    # _wrap_transform_fast_result
+                    result = getattr(self, func)(*args, **kwargs)
 
             return self._wrap_transform_fast_result(result)
 

pandas/tests/groupby/transform/test_transform.py

@@ -1549,3 +1549,17 @@ def test_transform_aligns_depr(func, series, expected_values, keys, keys_in_inde
         if series:
             expected = expected["b"]
         tm.assert_equal(result, expected)
+
+
+@pytest.mark.parametrize("keys", ["A", ["A", "B"]])
+def test_as_index_no_change(keys, df, groupby_func):
+    # GH#49834 - as_index should have no impact on DataFrameGroupBy.transform
+    if keys == "A":
+        # Column B is string dtype; will fail on some ops
+        df = df.drop(columns="B")
+    args = get_groupby_method_args(groupby_func, df)
+    gb_as_index_true = df.groupby(keys, as_index=True)
+    gb_as_index_false = df.groupby(keys, as_index=False)
+    result = gb_as_index_true.transform(groupby_func, *args)
+    expected = gb_as_index_false.transform(groupby_func, *args)
+    tm.assert_equal(result, expected)