BUG: fixed issue with mixed type groupby aggregate

Fixes issue #16916, where using aggregate on a mixed type grouping
vector fails. Added test in test_aggregate.py to ensure that the bug
is fixed.

Author: Nicolai Reeve

pandas/core/groupby.py

@@ -2883,7 +2883,12 @@ def aggregate(self, func_or_funcs, *args, **kwargs):
             except Exception:
                 result = self._aggregate_named(func_or_funcs, *args, **kwargs)
 
-            index = Index(sorted(result), name=self.grouper.names[0])
+            # mixed types fail to sort
+            try:
+                values = sorted(result)
+            except TypeError:
+                values = result
+            index = Index(values, name=self.grouper.names[0])
             ret = Series(result, index=index)
 
         if not self.as_index:  # pragma: no cover

pandas/tests/groupby/test_aggregate.py

@@ -892,3 +892,15 @@ def test_sum_uint64_overflow(self):
         expected.index.name = 0
         result = df.groupby(0).sum()
         tm.assert_frame_equal(result, expected)
+
+    def test_mixed_type_grouping(self):
+        X = pd.DataFrame(data=[[[1, 1], [2, 2], [3, 3]],
+                               [[1, 1], [2, 2], [3, 3]]],
+                         columns=['X', 'Y', 'Z'],
+                         index=pd.Index(data=[2, 'g1'], name='grouping'))
+
+        S = pd.DataFrame(data=[[1, 2, 3], [1, 2, 3], [1, 2, 3], [1, 2, 3]],
+                         columns=list('XYZ'), index=list('qwer'))
+        S['grouping'] = ['g1', 'g1', 2, 2]
+        T = S.groupby('grouping').aggregate(lambda x: x.tolist())
+        tm.assert_frame_equal(T, X)