BUG: Fix #10355, std() groupby calculation

doc/source/whatsnew/v0.25.0.rst

@@ -442,6 +442,7 @@ Groupby/Resample/Rolling
 - Bug in :meth:`pandas.core.groupby.GroupBy.idxmax` and :meth:`pandas.core.groupby.GroupBy.idxmin` with datetime column would return incorrect dtype (:issue:`25444`, :issue:`15306`)
 - Bug in :meth:`pandas.core.groupby.GroupBy.cumsum`, :meth:`pandas.core.groupby.GroupBy.cumprod`, :meth:`pandas.core.groupby.GroupBy.cummin` and :meth:`pandas.core.groupby.GroupBy.cummax` with categorical column having absent categories, would return incorrect result or segfault (:issue:`16771`)
 - Bug in :meth:`pandas.core.groupby.GroupBy.nth` where NA values in the grouping would return incorrect results (:issue:`26011`)
+- Bug in :meth:`pandas.core.groupby.GroupBy.std` that computed standard deviation without respecting groupby context when ``as_index=False`` (:issue:`10355`)
 - Bug in :meth:`pandas.core.groupby.SeriesGroupBy.transform` where transforming an empty group would raise error (:issue:`26208`)
 
 

pandas/_libs/reduction.pyx

@@ -238,12 +238,15 @@ cdef class SeriesBinGrouper:
         counts = np.zeros(self.ngroups, dtype=np.int64)
 
         if self.ngroups > 0:
-            counts[0] = self.bins[0]
-            for i in range(1, self.ngroups):
-                if i == self.ngroups - 1:
-                    counts[i] = len(self.arr) - self.bins[i - 1]
-                else:
-                    counts[i] = self.bins[i] - self.bins[i - 1]
+            if len(self.bins) == 0:
+                return np.empty(0, dtype='O'), counts
+            else:
+                counts[0] = self.bins[0]
+                for i in range(1, self.ngroups):
+                    if i == self.ngroups - 1:
+                        counts[i] = len(self.arr) - self.bins[i - 1]
+                    else:
+                        counts[i] = self.bins[i] - self.bins[i - 1]
 
         group_size = 0
         n = len(self.arr)

pandas/core/groupby/groupby.py

@@ -1207,10 +1207,16 @@ def std(self, ddof=1, *args, **kwargs):
         Series or DataFrame
             Standard deviation of values within each group.
         """
-
-        # TODO: implement at Cython level?
         nv.validate_groupby_func('std', args, kwargs)
-        return np.sqrt(self.var(ddof=ddof, **kwargs))
+        if ddof == 1:
+            try:
+                return self._cython_agg_general('std', **kwargs)
+            except Exception:
+                pass
+
+        f = lambda x: x.std(ddof=ddof, **kwargs)
+        with _group_selection_context(self):
+            return self._python_agg_general(f)
 
     @Substitution(name='groupby')
     @Appender(_common_see_also)
@@ -1235,13 +1241,11 @@ def var(self, ddof=1, *args, **kwargs):
             try:
                 return self._cython_agg_general('var', **kwargs)
             except Exception:
-                f = lambda x: x.var(ddof=ddof, **kwargs)
-                with _group_selection_context(self):
-                    return self._python_agg_general(f)
-        else:
-            f = lambda x: x.var(ddof=ddof, **kwargs)
-            with _group_selection_context(self):
-                return self._python_agg_general(f)
+                pass
+
+        f = lambda x: x.var(ddof=ddof, **kwargs)
+        with _group_selection_context(self):
+            return self._python_agg_general(f)
 
     @Substitution(name='groupby')
     @Appender(_common_see_also)

pandas/core/groupby/ops.py

@@ -345,6 +345,10 @@ def get_group_levels(self):
                 'name': 'group_median'
             },
             'var': 'group_var',
+            'std': {
+                'name': 'group_var_bin',
+                'f': lambda func, a: np.sqrt(func(a)),
+            },
             'first': {
                 'name': 'group_nth',
                 'f': lambda func, a, b, c, d, e: func(a, b, c, d, 1, -1)

pandas/tests/groupby/test_whitelist.py

@@ -161,36 +161,48 @@ def raw_frame():
 
 @pytest.mark.parametrize('op', AGG_FUNCTIONS)
 @pytest.mark.parametrize('level', [0, 1])
-@pytest.mark.parametrize('axis', [0, 1])
 @pytest.mark.parametrize('skipna', [True, False])
 @pytest.mark.parametrize('sort', [True, False])
+@pytest.mark.parametrize('as_index', [True, False])
 def test_regression_whitelist_methods(
         raw_frame, op, level,
-        axis, skipna, sort):
+        axis, skipna, sort, as_index):
     # GH6944
     # GH 17537
     # explicitly test the whitelist methods
 
-    if axis == 0:
+    if not as_index and axis not in [0, 'index']:
+        pytest.skip('as_index=False only valid for axis=0')
+
+    if axis in [0, 'index']:
         frame = raw_frame
     else:
         frame = raw_frame.T
 
+    if not isinstance(frame.index, MultiIndex) and (level > 0 or level < -1):
+        pytest.skip('level > 0 or level < -1 only valid with MultiIndex')
+
+    grouped = frame.groupby(level=level, axis=axis, sort=sort,
+                            as_index=as_index)
+
     if op in AGG_FUNCTIONS_WITH_SKIPNA:
-        grouped = frame.groupby(level=level, axis=axis, sort=sort)
         result = getattr(grouped, op)(skipna=skipna)
-        expected = getattr(frame, op)(level=level, axis=axis,
-                                      skipna=skipna)
-        if sort:
-            expected = expected.sort_index(axis=axis, level=level)
-        tm.assert_frame_equal(result, expected)
+        expected = getattr(frame, op)(level=level, axis=axis, skipna=skipna)
     else:
-        grouped = frame.groupby(level=level, axis=axis, sort=sort)
         result = getattr(grouped, op)()
         expected = getattr(frame, op)(level=level, axis=axis)
-        if sort:
-            expected = expected.sort_index(axis=axis, level=level)
-        tm.assert_frame_equal(result, expected)
+
+    if sort:
+        expected = expected.sort_index(axis=axis, level=level)
+
+    if not as_index:
+        expected = expected.reset_index()
+        if level == 0:
+            expected = expected.drop(columns=['first'])
+        if level == 1:
+            expected = expected.drop(columns=['second'])
+
+    tm.assert_frame_equal(result, expected)
 
 
 def test_groupby_blacklist(df_letters):