BUG: Validate numeric_only parameter in groupby aggregations

doc/source/whatsnew/v3.0.0.rst

@@ -1150,6 +1150,7 @@ Groupby/resample/rolling
 - Bug in :meth:`.DataFrameGroupBy.groups` and :meth:`.SeriesGroupBy.groups` would fail when the groups were :class:`Categorical` with an NA value (:issue:`61356`)
 - Bug in :meth:`.DataFrameGroupBy.groups` and :meth:`.SeriesGroupby.groups` that would not respect groupby argument ``dropna`` (:issue:`55919`)
 - Bug in :meth:`.DataFrameGroupBy.median` where nat values gave an incorrect result. (:issue:`57926`)
+- Bug in :meth:`.DataFrameGroupBy` reductions where boolean-valued inputs were mishandled in the Cython aggregation path (``_cython_agg_general``); adding an ``is_bool`` check fixes incorrect results for some bool inputs. (:issue:`62778`)
 - Bug in :meth:`.DataFrameGroupBy.quantile` when ``interpolation="nearest"`` is inconsistent with :meth:`DataFrame.quantile` (:issue:`47942`)
 - Bug in :meth:`.Resampler.interpolate` on a :class:`DataFrame` with non-uniform sampling and/or indices not aligning with the resulting resampled index would result in wrong interpolation (:issue:`21351`)
 - Bug in :meth:`.Series.rolling` when used with a :class:`.BaseIndexer` subclass and computing min/max (:issue:`46726`)

pandas/core/groupby/groupby.py

@@ -86,6 +86,7 @@ class providing the base-class of operations.
     is_object_dtype,
     is_scalar,
     is_string_dtype,
+    is_bool,
     needs_i8_conversion,
     pandas_dtype,
 )
@@ -1754,7 +1755,10 @@ def _cython_agg_general(
         **kwargs,
     ):
         # Note: we never get here with how="ohlc" for DataFrameGroupBy;
-        #  that goes through SeriesGroupBy
+        # that goes through SeriesGroupBy
+
+        if not is_bool(numeric_only):
+            raise ValueError("numeric_only accepts only Boolean values")
 
         data = self._get_data_to_aggregate(numeric_only=numeric_only, name=how)
 

pandas/tests/groupby/test_reductions.py

@@ -1520,3 +1520,27 @@ def test_groupby_std_datetimelike():
     exp_ser = Series([td1 * 2, td1, td1, td1, td4], index=np.arange(5))
     expected = DataFrame({"A": exp_ser, "B": exp_ser, "C": exp_ser})
     tm.assert_frame_equal(result, expected)
+
+def test_mean_numeric_only_validates_bool():
+    """
+    Test that numeric_only parameter only accepts boolean values.
+    See GH#62778
+    """
+    df = pd.DataFrame({"A": range(5), "B": range(5)})
+
+    # These test cases should raise a ValueError
+    msg = "numeric_only accepts only Boolean values"
+    with pytest.raises(ValueError, match=msg):
+        df.groupby(["A"]).mean(["B"])
+
+    with pytest.raises(ValueError, match=msg):
+        df.groupby(["A"]).mean(numeric_only="True")
+
+    with pytest.raises(ValueError, match=msg):
+        df.groupby(["A"]).mean(numeric_only=1)
+
+    # These test cases should work absolutely fine
+    df.groupby(["A"]).mean()
+    df.groupby(["A"]).mean(numeric_only=True)
+    df.groupby(["A"]).mean(numeric_only=False)
+