BUG: Fix metadata propagation in reductions (#53542)

* BUG: Fix metadata propagation in reductions

* fix tests

* actually fix tests

* fix typing

* Update pandas/core/reshape/encoding.py

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Co-authored-by: Matthew Roeschke <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: 3 people

doc/source/whatsnew/v2.1.0.rst

@@ -772,8 +772,10 @@ Styler
 
 Metadata
 ^^^^^^^^
+- Fixed metadata propagation in :meth:`DataFrame.max`, :meth:`DataFrame.min`, :meth:`DataFrame.prod`, :meth:`DataFrame.mean`, :meth:`Series.mode`, :meth:`DataFrame.median`, :meth:`DataFrame.sem`, :meth:`DataFrame.skew`, :meth:`DataFrame.kurt` (:issue:`28283`)
 - Fixed metadata propagation in :meth:`DataFrame.squeeze`, and :meth:`DataFrame.describe` (:issue:`28283`)
 - Fixed metadata propagation in :meth:`DataFrame.std` (:issue:`28283`)
+-
 
 Other
 ^^^^^

pandas/core/frame.py

@@ -11137,12 +11137,13 @@ def any(  # type: ignore[override]
         bool_only: bool = False,
         skipna: bool = True,
         **kwargs,
-    ) -> Series:
-        # error: Incompatible return value type (got "Union[Series, bool]",
-        # expected "Series")
-        return self._logical_func(  # type: ignore[return-value]
+    ) -> Series | bool:
+        result = self._logical_func(
             "any", nanops.nanany, axis, bool_only, skipna, **kwargs
         )
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="any")
+        return result
 
     @doc(make_doc("all", ndim=2))
     def all(
@@ -11151,12 +11152,13 @@ def all(
         bool_only: bool = False,
         skipna: bool = True,
         **kwargs,
-    ) -> Series:
-        # error: Incompatible return value type (got "Union[Series, bool]",
-        # expected "Series")
-        return self._logical_func(  # type: ignore[return-value]
+    ) -> Series | bool:
+        result = self._logical_func(
             "all", nanops.nanall, axis, bool_only, skipna, **kwargs
         )
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="all")
+        return result
 
     @doc(make_doc("min", ndim=2))
     def min(
@@ -11166,7 +11168,10 @@ def min(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().min(axis, skipna, numeric_only, **kwargs)
+        result = super().min(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="min")
+        return result
 
     @doc(make_doc("max", ndim=2))
     def max(
@@ -11176,7 +11181,10 @@ def max(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().max(axis, skipna, numeric_only, **kwargs)
+        result = super().max(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="max")
+        return result
 
     @doc(make_doc("sum", ndim=2))
     def sum(
@@ -11199,7 +11207,8 @@ def prod(
         min_count: int = 0,
         **kwargs,
     ):
-        return super().prod(axis, skipna, numeric_only, min_count, **kwargs)
+        result = super().prod(axis, skipna, numeric_only, min_count, **kwargs)
+        return result.__finalize__(self, method="prod")
 
     @doc(make_doc("mean", ndim=2))
     def mean(
@@ -11209,7 +11218,10 @@ def mean(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().mean(axis, skipna, numeric_only, **kwargs)
+        result = super().mean(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="mean")
+        return result
 
     @doc(make_doc("median", ndim=2))
     def median(
@@ -11219,7 +11231,10 @@ def median(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().median(axis, skipna, numeric_only, **kwargs)
+        result = super().median(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="median")
+        return result
 
     @doc(make_doc("sem", ndim=2))
     def sem(
@@ -11230,7 +11245,10 @@ def sem(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().sem(axis, skipna, ddof, numeric_only, **kwargs)
+        result = super().sem(axis, skipna, ddof, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="sem")
+        return result
 
     @doc(make_doc("var", ndim=2))
     def var(
@@ -11241,7 +11259,10 @@ def var(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().var(axis, skipna, ddof, numeric_only, **kwargs)
+        result = super().var(axis, skipna, ddof, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="var")
+        return result
 
     @doc(make_doc("std", ndim=2))
     def std(
@@ -11252,8 +11273,10 @@ def std(
         numeric_only: bool = False,
         **kwargs,
     ):
-        result = cast(Series, super().std(axis, skipna, ddof, numeric_only, **kwargs))
-        return result.__finalize__(self, method="std")
+        result = super().std(axis, skipna, ddof, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="std")
+        return result
 
     @doc(make_doc("skew", ndim=2))
     def skew(
@@ -11263,7 +11286,10 @@ def skew(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().skew(axis, skipna, numeric_only, **kwargs)
+        result = super().skew(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="skew")
+        return result
 
     @doc(make_doc("kurt", ndim=2))
     def kurt(
@@ -11273,7 +11299,10 @@ def kurt(
         numeric_only: bool = False,
         **kwargs,
     ):
-        return super().kurt(axis, skipna, numeric_only, **kwargs)
+        result = super().kurt(axis, skipna, numeric_only, **kwargs)
+        if isinstance(result, Series):
+            result = result.__finalize__(self, method="kurt")
+        return result
 
     kurtosis = kurt
     product = prod

pandas/core/reshape/encoding.py

@@ -6,7 +6,10 @@
     Iterable,
 )
 import itertools
-from typing import TYPE_CHECKING
+from typing import (
+    TYPE_CHECKING,
+    cast,
+)
 
 import numpy as np
 
@@ -455,10 +458,12 @@ def from_dummies(
             f"Received 'data' of type: {type(data).__name__}"
         )
 
-    if data.isna().any().any():
+    col_isna_mask = cast(Series, data.isna().any())
+
+    if col_isna_mask.any():
         raise ValueError(
             "Dummy DataFrame contains NA value in column: "
-            f"'{data.isna().any().idxmax()}'"
+            f"'{col_isna_mask.idxmax()}'"
         )
 
     # index data with a list of all columns that are dummies

pandas/core/series.py

@@ -2202,7 +2202,7 @@ def mode(self, dropna: bool = True) -> Series:
         # Ensure index is type stable (should always use int index)
         return self._constructor(
             res_values, index=range(len(res_values)), name=self.name, copy=False
-        )
+        ).__finalize__(self, method="mode")
 
     def unique(self) -> ArrayLike:  # pylint: disable=useless-parent-delegation
         """

pandas/tests/generic/test_finalize.py

@@ -180,10 +180,8 @@
     (pd.DataFrame, frame_data, operator.methodcaller("idxmin")),
     (pd.DataFrame, frame_data, operator.methodcaller("idxmax")),
     (pd.DataFrame, frame_data, operator.methodcaller("mode")),
-    pytest.param(
-        (pd.Series, [0], operator.methodcaller("mode")),
-        marks=not_implemented_mark,
-    ),
+    (pd.Series, [0], operator.methodcaller("mode")),
+    (pd.DataFrame, frame_data, operator.methodcaller("median")),
     (
         pd.DataFrame,
         frame_data,
@@ -363,17 +361,24 @@
     # Cumulative reductions
     (pd.Series, ([1],), operator.methodcaller("cumsum")),
     (pd.DataFrame, frame_data, operator.methodcaller("cumsum")),
+    (pd.Series, ([1],), operator.methodcaller("cummin")),
+    (pd.DataFrame, frame_data, operator.methodcaller("cummin")),
+    (pd.Series, ([1],), operator.methodcaller("cummax")),
+    (pd.DataFrame, frame_data, operator.methodcaller("cummax")),
+    (pd.Series, ([1],), operator.methodcaller("cumprod")),
+    (pd.DataFrame, frame_data, operator.methodcaller("cumprod")),
     # Reductions
-    pytest.param(
-        (pd.DataFrame, frame_data, operator.methodcaller("any")),
-        marks=not_implemented_mark,
-    ),
+    (pd.DataFrame, frame_data, operator.methodcaller("any")),
+    (pd.DataFrame, frame_data, operator.methodcaller("all")),
+    (pd.DataFrame, frame_data, operator.methodcaller("min")),
+    (pd.DataFrame, frame_data, operator.methodcaller("max")),
     (pd.DataFrame, frame_data, operator.methodcaller("sum")),
     (pd.DataFrame, frame_data, operator.methodcaller("std")),
-    pytest.param(
-        (pd.DataFrame, frame_data, operator.methodcaller("mean")),
-        marks=not_implemented_mark,
-    ),
+    (pd.DataFrame, frame_data, operator.methodcaller("mean")),
+    (pd.DataFrame, frame_data, operator.methodcaller("prod")),
+    (pd.DataFrame, frame_data, operator.methodcaller("sem")),
+    (pd.DataFrame, frame_data, operator.methodcaller("skew")),
+    (pd.DataFrame, frame_data, operator.methodcaller("kurt")),
 ]