This fix enables you to preserve the datetime precision when using the melt method. It fixes the issues raised in #55254.

Author: paulreece

doc/source/whatsnew/v2.2.0.rst

@@ -77,7 +77,6 @@ Other enhancements
 - :func:`read_csv` now supports ``on_bad_lines`` parameter with ``engine="pyarrow"``. (:issue:`54480`)
 - :meth:`ExtensionArray._explode` interface method added to allow extension type implementations of the ``explode`` method (:issue:`54833`)
 - DataFrame.apply now allows the usage of numba (via ``engine="numba"``) to JIT compile the passed function, allowing for potential speedups (:issue:`54666`)
-- Implement masked algorithms for :meth:`Series.value_counts` (:issue:`54984`)
 -
 
 .. ---------------------------------------------------------------------------
@@ -219,7 +218,6 @@ Other Deprecations
 - Deprecated allowing non-keyword arguments in :meth:`DataFrame.to_parquet` except ``path``. (:issue:`54229`)
 - Deprecated allowing non-keyword arguments in :meth:`DataFrame.to_pickle` except ``path``. (:issue:`54229`)
 - Deprecated allowing non-keyword arguments in :meth:`DataFrame.to_string` except ``buf``. (:issue:`54229`)
-- Deprecated allowing non-keyword arguments in :meth:`DataFrame.to_xml` except ``path_or_buffer``. (:issue:`54229`)
 - Deprecated automatic downcasting of object-dtype results in :meth:`Series.replace` and :meth:`DataFrame.replace`, explicitly call ``result = result.infer_objects(copy=False)`` instead. To opt in to the future version, use ``pd.set_option("future.no_silent_downcasting", True)`` (:issue:`54710`)
 - Deprecated downcasting behavior in :meth:`Series.where`, :meth:`DataFrame.where`, :meth:`Series.mask`, :meth:`DataFrame.mask`, :meth:`Series.clip`, :meth:`DataFrame.clip`; in a future version these will not infer object-dtype columns to non-object dtype, or all-round floats to integer dtype. Call ``result.infer_objects(copy=False)`` on the result for object inference, or explicitly cast floats to ints. To opt in to the future version, use ``pd.set_option("future.no_silent_downcasting", True)`` (:issue:`53656`)
 - Deprecated including the groups in computations when using :meth:`DataFrameGroupBy.apply` and :meth:`DataFrameGroupBy.resample`; pass ``include_groups=False`` to exclude the groups (:issue:`7155`)
@@ -252,6 +250,7 @@ Bug fixes
 - Bug in :class:`AbstractHolidayCalendar` where timezone data was not propagated when computing holiday observances (:issue:`54580`)
 - Bug in :class:`pandas.core.window.Rolling` where duplicate datetimelike indexes are treated as consecutive rather than equal with ``closed='left'`` and ``closed='neither'`` (:issue:`20712`)
 - Bug in :meth:`DataFrame.apply` where passing ``raw=True`` ignored ``args`` passed to the applied function (:issue:`55009`)
+- Bug in :meth:`pandas.DataFrame.melt` where it would not preserve the datetime (:issue:`55254`)
 - Bug in :meth:`pandas.read_excel` with a ODS file without cached formatted cell for float values (:issue:`55219`)
 
 Categorical
@@ -316,7 +315,6 @@ MultiIndex
 I/O
 ^^^
 - Bug in :func:`read_csv` where ``on_bad_lines="warn"`` would write to ``stderr`` instead of raise a Python warning. This now yields a :class:`.errors.ParserWarning` (:issue:`54296`)
-- Bug in :func:`read_csv` with ``engine="pyarrow"`` where ``usecols`` wasn't working with a csv with no headers (:issue:`54459`)
 - Bug in :func:`read_excel`, with ``engine="xlrd"`` (``xls`` files) erroring when file contains NaNs/Infs (:issue:`54564`)
 - Bug in :func:`to_excel`, with ``OdsWriter`` (``ods`` files) writing boolean/string value (:issue:`54994`)
 

pandas/core/reshape/melt.py

@@ -134,7 +134,9 @@ def melt(
 
     mcolumns = id_vars + var_name + [value_name]
 
-    if frame.shape[1] > 0:
+    if frame.shape[1] > 0 and not any(
+        not isinstance(dt, np.dtype) and dt._supports_2d for dt in frame.dtypes.values
+    ):
         mdata[value_name] = concat(
             [frame.iloc[:, i] for i in range(frame.shape[1])]
         ).values

pandas/tests/reshape/test_melt.py

@@ -459,6 +459,47 @@ def test_melt_ea_columns(self):
         )
         tm.assert_frame_equal(result, expected)
 
+    def test_melt_preserves_datetime(self):
+        df = DataFrame(
+            data=[
+                {
+                    "type": "A0",
+                    "start_date": pd.Timestamp("2023/03/01", tz="Asia/Tokyo"),
+                    "end_date": pd.Timestamp("2023/03/10", tz="Asia/Tokyo"),
+                },
+                {
+                    "type": "A1",
+                    "start_date": pd.Timestamp("2023/03/01", tz="Asia/Tokyo"),
+                    "end_date": pd.Timestamp("2023/03/11", tz="Asia/Tokyo"),
+                },
+            ],
+            index=["aaaa", "bbbb"],
+        )
+        result = df.melt(
+            id_vars=["type"],
+            value_vars=["start_date", "end_date"],
+            var_name="start/end",
+            value_name="date",
+        )
+        expected = DataFrame(
+            {
+                "type": {0: "A0", 1: "A1", 2: "A0", 3: "A1"},
+                "start/end": {
+                    0: "start_date",
+                    1: "start_date",
+                    2: "end_date",
+                    3: "end_date",
+                },
+                "date": {
+                    0: pd.Timestamp("2023-03-01 00:00:00+0900", tz="Asia/Tokyo"),
+                    1: pd.Timestamp("2023-03-01 00:00:00+0900", tz="Asia/Tokyo"),
+                    2: pd.Timestamp("2023-03-10 00:00:00+0900", tz="Asia/Tokyo"),
+                    3: pd.Timestamp("2023-03-11 00:00:00+0900", tz="Asia/Tokyo"),
+                },
+            }
+        )
+        tm.assert_frame_equal(result, expected)
+
 
 class TestLreshape:
     def test_pairs(self):