Add keyword sort to pivot_table (#40954)

Author: phofl

doc/source/whatsnew/v1.3.0.rst

@@ -221,6 +221,7 @@ Other enhancements
 - :meth:`pandas.read_csv` and :meth:`pandas.read_json` expose the argument ``encoding_errors`` to control how encoding errors are handled (:issue:`39450`)
 - :meth:`.GroupBy.any` and :meth:`.GroupBy.all` use Kleene logic with nullable data types (:issue:`37506`)
 - :meth:`.GroupBy.any` and :meth:`.GroupBy.all` return a ``BooleanDtype`` for columns with nullable data types (:issue:`33449`)
+- Add keyword ``sort`` to :func:`pivot_table` to allow non-sorting of the result (:issue:`39143`)
 -
 
 .. ---------------------------------------------------------------------------

pandas/core/frame.py

@@ -7781,6 +7781,11 @@ def pivot(self, index=None, columns=None, values=None) -> DataFrame:
 
             .. versionchanged:: 0.25.0
 
+        sort : bool, default True
+            Specifies if the result should be sorted.
+
+            .. versionadded:: 1.3.0
+
         Returns
         -------
         DataFrame
@@ -7884,6 +7889,7 @@ def pivot_table(
         dropna=True,
         margins_name="All",
         observed=False,
+        sort=True,
     ) -> DataFrame:
         from pandas.core.reshape.pivot import pivot_table
 
@@ -7898,6 +7904,7 @@ def pivot_table(
             dropna=dropna,
             margins_name=margins_name,
             observed=observed,
+            sort=sort,
         )
 
     def stack(self, level: Level = -1, dropna: bool = True):

pandas/core/reshape/pivot.py

@@ -64,6 +64,7 @@ def pivot_table(
     dropna=True,
     margins_name="All",
     observed=False,
+    sort=True,
 ) -> DataFrame:
     index = _convert_by(index)
     columns = _convert_by(columns)
@@ -83,6 +84,7 @@ def pivot_table(
                 dropna=dropna,
                 margins_name=margins_name,
                 observed=observed,
+                sort=sort,
             )
             pieces.append(_table)
             keys.append(getattr(func, "__name__", func))
@@ -101,6 +103,7 @@ def pivot_table(
         dropna,
         margins_name,
         observed,
+        sort,
     )
     return table.__finalize__(data, method="pivot_table")
 
@@ -116,6 +119,7 @@ def __internal_pivot_table(
     dropna: bool,
     margins_name: str,
     observed: bool,
+    sort: bool,
 ) -> DataFrame:
     """
     Helper of :func:`pandas.pivot_table` for any non-list ``aggfunc``.
@@ -157,7 +161,7 @@ def __internal_pivot_table(
                 pass
         values = list(values)
 
-    grouped = data.groupby(keys, observed=observed)
+    grouped = data.groupby(keys, observed=observed, sort=sort)
     agged = grouped.agg(aggfunc)
     if dropna and isinstance(agged, ABCDataFrame) and len(agged.columns):
         agged = agged.dropna(how="all")

pandas/tests/reshape/test_pivot.py

@@ -2115,6 +2115,28 @@ def test_pivot_table_doctest_case(self):
         expected = DataFrame(vals, columns=cols, index=index)
         tm.assert_frame_equal(table, expected)
 
+    def test_pivot_table_sort_false(self):
+        # GH#39143
+        df = DataFrame(
+            {
+                "a": ["d1", "d4", "d3"],
+                "col": ["a", "b", "c"],
+                "num": [23, 21, 34],
+                "year": ["2018", "2018", "2019"],
+            }
+        )
+        result = df.pivot_table(
+            index=["a", "col"], columns="year", values="num", aggfunc="sum", sort=False
+        )
+        expected = DataFrame(
+            [[23, np.nan], [21, np.nan], [np.nan, 34]],
+            columns=Index(["2018", "2019"], name="year"),
+            index=MultiIndex.from_arrays(
+                [["d1", "d4", "d3"], ["a", "b", "c"]], names=["a", "col"]
+            ),
+        )
+        tm.assert_frame_equal(result, expected)
+
 
 class TestPivot:
     def test_pivot(self):