BUG: DataFrame.min/max dt64 with skipna=False (pandas-dev#37425)

doc/source/whatsnew/v1.2.0.rst

@@ -404,6 +404,7 @@ Numeric
 - Bug in :class:`IntervalArray` comparisons with :class:`Series` not returning :class:`Series` (:issue:`36908`)
 - Bug in :class:`DataFrame` allowing arithmetic operations with list of array-likes with undefined results. Behavior changed to raising ``ValueError`` (:issue:`36702`)
 - Bug in :meth:`DataFrame.std`` with ``timedelta64`` dtype and ``skipna=False`` (:issue:`37392`)
+- Bug in :meth:`DataFrame.min` and :meth:`DataFrame.max` with ``datetime64`` dtype and ``skipna=False`` (:issue:`36907`)
 
 Conversion
 ^^^^^^^^^^

pandas/compat/numpy/function.py

@@ -387,20 +387,20 @@ def validate_resampler_func(method: str, args, kwargs) -> None:
             raise TypeError("too many arguments passed in")
 
 
-def validate_minmax_axis(axis: Optional[int]) -> None:
+def validate_minmax_axis(axis: Optional[int], ndim: int = 1) -> None:
     """
     Ensure that the axis argument passed to min, max, argmin, or argmax is zero
     or None, as otherwise it will be incorrectly ignored.
 
     Parameters
     ----------
     axis : int or None
+    ndim : int, default 1
 
     Raises
     ------
     ValueError
     """
-    ndim = 1  # hard-coded for Index
     if axis is None:
         return
     if axis >= ndim or (axis < 0 and ndim + axis < 0):

pandas/core/arrays/datetimelike.py

@@ -1264,13 +1264,24 @@ def min(self, axis=None, skipna=True, *args, **kwargs):
         Series.min : Return the minimum value in a Series.
         """
         nv.validate_min(args, kwargs)
-        nv.validate_minmax_axis(axis)
+        nv.validate_minmax_axis(axis, self.ndim)
 
-        result = nanops.nanmin(self.asi8, skipna=skipna, mask=self.isna())
-        if isna(result):
-            # Period._from_ordinal does not handle np.nan gracefully
-            return NaT
-        return self._box_func(result)
+        if is_period_dtype(self.dtype):
+            # pass datetime64 values to nanops to get correct NaT semantics
+            result = nanops.nanmin(
+                self._ndarray.view("M8[ns]"), axis=axis, skipna=skipna
+            )
+            if result is NaT:
+                return NaT
+            result = result.view("i8")
+            if axis is None or self.ndim == 1:
+                return self._box_func(result)
+            return self._from_backing_data(result)
+
+        result = nanops.nanmin(self._ndarray, axis=axis, skipna=skipna)
+        if lib.is_scalar(result):
+            return self._box_func(result)
+        return self._from_backing_data(result)
 
     def max(self, axis=None, skipna=True, *args, **kwargs):
         """
@@ -1286,23 +1297,24 @@ def max(self, axis=None, skipna=True, *args, **kwargs):
         # TODO: skipna is broken with max.
         # See https://github.com/pandas-dev/pandas/issues/24265
         nv.validate_max(args, kwargs)
-        nv.validate_minmax_axis(axis)
-
-        mask = self.isna()
-        if skipna:
-            values = self[~mask].asi8
-        elif mask.any():
-            return NaT
-        else:
-            values = self.asi8
+        nv.validate_minmax_axis(axis, self.ndim)
 
-        if not len(values):
-            # short-circuit for empty max / min
-            return NaT
+        if is_period_dtype(self.dtype):
+            # pass datetime64 values to nanops to get correct NaT semantics
+            result = nanops.nanmax(
+                self._ndarray.view("M8[ns]"), axis=axis, skipna=skipna
+            )
+            if result is NaT:
+                return result
+            result = result.view("i8")
+            if axis is None or self.ndim == 1:
+                return self._box_func(result)
+            return self._from_backing_data(result)
 
-        result = nanops.nanmax(values, skipna=skipna)
-        # Don't have to worry about NA `result`, since no NA went in.
-        return self._box_func(result)
+        result = nanops.nanmax(self._ndarray, axis=axis, skipna=skipna)
+        if lib.is_scalar(result):
+            return self._box_func(result)
+        return self._from_backing_data(result)
 
     def mean(self, skipna=True, axis: Optional[int] = 0):
         """

pandas/core/nanops.py

@@ -340,6 +340,7 @@ def _wrap_results(result, dtype: DtypeObj, fill_value=None):
             if result == fill_value:
                 result = np.nan
             if tz is not None:
+                # we get here e.g. via nanmean when we call it on a DTA[tz]
                 result = Timestamp(result, tz=tz)
             elif isna(result):
                 result = np.datetime64("NaT", "ns")
@@ -919,10 +920,13 @@ def reduction(
         mask: Optional[np.ndarray] = None,
     ) -> Dtype:
 
+        orig_values = values
         values, mask, dtype, dtype_max, fill_value = _get_values(
             values, skipna, fill_value_typ=fill_value_typ, mask=mask
         )
 
+        datetimelike = orig_values.dtype.kind in ["m", "M"]
+
         if (axis is not None and values.shape[axis] == 0) or values.size == 0:
             try:
                 result = getattr(values, meth)(axis, dtype=dtype_max)
@@ -933,7 +937,12 @@ def reduction(
             result = getattr(values, meth)(axis)
 
         result = _wrap_results(result, dtype, fill_value)
-        return _maybe_null_out(result, axis, mask, values.shape)
+        result = _maybe_null_out(result, axis, mask, values.shape)
+
+        if datetimelike and not skipna:
+            result = _mask_datetimelike_result(result, axis, mask, orig_values)
+
+        return result
 
     return reduction
 

pandas/tests/frame/test_analytics.py

@@ -1169,6 +1169,31 @@ def test_min_max_dt64_with_NaT(self):
         exp = Series([pd.NaT], index=["foo"])
         tm.assert_series_equal(res, exp)
 
+    def test_min_max_dt64_with_NaT_skipna_false(self, tz_naive_fixture):
+        # GH#36907
+        tz = tz_naive_fixture
+        df = DataFrame(
+            {
+                "a": [
+                    Timestamp("2020-01-01 08:00:00", tz=tz),
+                    Timestamp("1920-02-01 09:00:00", tz=tz),
+                ],
+                "b": [Timestamp("2020-02-01 08:00:00", tz=tz), pd.NaT],
+            }
+        )
+
+        res = df.min(axis=1, skipna=False)
+        expected = Series([df.loc[0, "a"], pd.NaT])
+        assert expected.dtype == df["a"].dtype
+
+        tm.assert_series_equal(res, expected)
+
+        res = df.max(axis=1, skipna=False)
+        expected = Series([df.loc[0, "b"], pd.NaT])
+        assert expected.dtype == df["a"].dtype
+
+        tm.assert_series_equal(res, expected)
+
     def test_min_max_dt64_api_consistency_with_NaT(self):
         # Calling the following sum functions returned an error for dataframes but
         # returned NaT for series. These tests check that the API is consistent in