Merge pull request #3913 from jreback/groupby_apply

BUG: (GH3911) groupby appying with custom function not converting dtypes of result

Author: jreback

RELEASE.rst

@@ -234,8 +234,11 @@ pandas 0.11.1
   - ``read_html`` now correctly skips tests (GH3741_)
   - PandasObjects raise TypeError when trying to hash (GH3882_)
   - Fix incorrect arguments passed to concat that are not list-like (e.g. concat(df1,df2)) (GH3481_)
-  - Correctly parse when passed the ``dtype=str`` (or other variable-len string dtypes) in ``read_csv`` (GH3795_)
+  - Correctly parse when passed the ``dtype=str`` (or other variable-len string dtypes)
+    in ``read_csv`` (GH3795_)
   - Fix index name not propogating when using ``loc/ix`` (GH3880_)
+  - Fix groupby when applying a custom function resulting in a returned DataFrame was 
+    not converting dtypes (GH3911_)
 
 .. _GH3164: https://github.com/pydata/pandas/issues/3164
 .. _GH2786: https://github.com/pydata/pandas/issues/2786
@@ -331,6 +334,7 @@ pandas 0.11.1
 .. _GH3873: https://github.com/pydata/pandas/issues/3873
 .. _GH3877: https://github.com/pydata/pandas/issues/3877
 .. _GH3880: https://github.com/pydata/pandas/issues/3880
+.. _GH3911: https://github.com/pydata/pandas/issues/3911
 
 
 pandas 0.11.0

pandas/core/groupby.py

@@ -1928,7 +1928,7 @@ def _wrap_applied_output(self, keys, values, not_indexed_same=False):
                     return Series(values, index=key_index)
 
                 return DataFrame(stacked_values, index=index,
-                                 columns=columns)
+                                 columns=columns).convert_objects()
 
             else:
                 return Series(values, index=key_index)

pandas/tests/test_groupby.py

@@ -261,6 +261,20 @@ def test_groupby_nonobject_dtype(self):
         expected = self.mframe.groupby(key.astype('O')).sum()
         assert_frame_equal(result, expected)
 
+        # GH 3911, mixed frame non-conversion
+        df = self.df_mixed_floats.copy()
+        df['value'] = range(len(df))
+
+        def max_value(group):
+            return group.ix[group['value'].idxmax()]
+
+        applied = df.groupby('A').apply(max_value)
+        result = applied.get_dtype_counts()
+        result.sort()
+        expected = Series({ 'object' : 2, 'float64' : 2, 'int64' : 1 })
+        expected.sort()
+        assert_series_equal(result,expected)
+
     def test_groupby_return_type(self):
 
         # GH2893, return a reduced type