ENH: add np.nan* funcs to cython_table

doc/source/whatsnew/v0.24.0.txt

@@ -1080,6 +1080,7 @@ Numeric
 - Bug in :meth:`DataFrame.astype` to extension dtype may raise ``AttributeError`` (:issue:`22578`)
 - Bug in :class:`DataFrame` with ``timedelta64[ns]`` dtype arithmetic operations with ``ndarray`` with integer dtype incorrectly treating the narray as ``timedelta64[ns]`` dtype (:issue:`23114`)
 - Bug in :meth:`Series.rpow` with object dtype ``NaN`` for ``1 ** NA`` instead of ``1`` (:issue:`22922`).
+- :meth:`Series.agg` can now handle numpy NaN-aware methods like :func:`numpy.nansum` (:issue:`19629`)
 
 Strings
 ^^^^^^^

pandas/core/base.py

@@ -178,27 +178,39 @@ class SelectionMixin(object):
     _selection = None
     _internal_names = ['_cache', '__setstate__']
     _internal_names_set = set(_internal_names)
+
     _builtin_table = OrderedDict((
         (builtins.sum, np.sum),
         (builtins.max, np.max),
         (builtins.min, np.min),
     ))
+
     _cython_table = OrderedDict((
         (builtins.sum, 'sum'),
         (builtins.max, 'max'),
         (builtins.min, 'min'),
         (np.all, 'all'),
         (np.any, 'any'),
         (np.sum, 'sum'),
+        (np.nansum, 'sum'),
         (np.mean, 'mean'),
+        (np.nanmean, 'mean'),
         (np.prod, 'prod'),
+        (np.nanprod, 'prod'),
         (np.std, 'std'),
+        (np.nanstd, 'std'),
         (np.var, 'var'),
+        (np.nanvar, 'var'),
         (np.median, 'median'),
+        (np.nanmedian, 'median'),
         (np.max, 'max'),
+        (np.nanmax, 'max'),
         (np.min, 'min'),
+        (np.nanmin, 'min'),
         (np.cumprod, 'cumprod'),
+        (np.nancumprod, 'cumprod'),
         (np.cumsum, 'cumsum'),
+        (np.nancumsum, 'cumsum'),
     ))
 
     @property

pandas/tests/groupby/aggregate/test_other.py

@@ -487,17 +487,6 @@ def test_agg_structs_series(structure, expected):
     tm.assert_series_equal(result, expected)
 
 
-@pytest.mark.parametrize('observed', [
-    True,
-    pytest.param(False,
-                 marks=pytest.mark.xfail(reason="GH#18869: agg func not "
-                                                "called on empty groups.",
-                                         strict=True)),
-    pytest.param(None,
-                 marks=pytest.mark.xfail(reason="GH#18869: agg func not "
-                                                "called on empty groups.",
-                                         strict=True))
-])
 def test_agg_category_nansum(observed):
     categories = ['a', 'b', 'c']
     df = pd.DataFrame({"A": pd.Categorical(['a', 'a', 'b'],