BUG: pivot_table sometimes returns Series (#4386)

Author: mroeschke

doc/source/whatsnew/v0.19.1.txt

@@ -80,4 +80,5 @@ Bug Fixes
 
 
 - Bug in ``pd.pivot_table`` may raise ``TypeError`` or ``ValueError`` when ``index`` or ``columns``
-  is not scalar and ``values`` is not specified (:issue:`14380`)
+  is not scalar and ``values`` is not specified (:issue:`14380`)
+- Bug in ``pd.pivot_table`` where a ``Series`` is return instead of a ``DataFrame`` (:issue:`4386`)

pandas/tests/test_categorical.py

@@ -3120,9 +3120,9 @@ def test_pivot_table(self):
             [Categorical(["a", "b", "z"], ordered=True),
              Categorical(["c", "d", "y"], ordered=True)],
             names=['A', 'B'])
-        expected = Series([1, 2, np.nan, 3, 4, np.nan, np.nan, np.nan, np.nan],
-                          index=exp_index, name='values')
-        tm.assert_series_equal(result, expected)
+        exp_data = [1, 2, np.nan, 3, 4, np.nan, np.nan, np.nan, np.nan]
+        expected = DataFrame(exp_data, index=exp_index, columns=['values'])
+        tm.assert_frame_equal(result, expected)
 
     def test_count(self):
 

pandas/tools/pivot.py

@@ -170,6 +170,9 @@ def pivot_table(data, values=None, index=None, columns=None, aggfunc='mean',
     if len(index) == 0 and len(columns) > 0:
         table = table.T
 
+    if isinstance(table, Series):
+        table = table.to_frame()
+
     return table
 
 

pandas/tools/tests/test_pivot.py

@@ -350,7 +350,7 @@ def _check_output(result, values_col, index=['A', 'B'],
         # no rows
         rtable = self.data.pivot_table(columns=['AA', 'BB'], margins=True,
                                        aggfunc=np.mean)
-        tm.assertIsInstance(rtable, Series)
+        tm.assertIsInstance(rtable, DataFrame)
 
         table = self.data.pivot_table(index=['AA', 'BB'], margins=True,
                                       aggfunc='mean')
@@ -485,7 +485,7 @@ def test_margins_no_values_no_cols(self):
         # Regression test on pivot table: no values or cols passed.
         result = self.data[['A', 'B']].pivot_table(
             index=['A', 'B'], aggfunc=len, margins=True)
-        result_list = result.tolist()
+        result_list = result.values.flatten().tolist()
         self.assertEqual(sum(result_list[:-1]), result_list[-1])
 
     def test_margins_no_values_two_rows(self):
@@ -854,6 +854,39 @@ def test_categorical_margins(self):
         table = data.pivot_table('x', 'y', 'z', margins=True)
         tm.assert_frame_equal(table, expected)
 
+    def test_always_return_dataframe(self):
+        # GH 4386
+        df = DataFrame({'col1': [3, 4, 5],
+                        'col2': ['C', 'D', 'E'],
+                        'col3': [1, 3, 9]})
+        result = df.pivot_table('col1', index=['col3', 'col2'], aggfunc=np.sum)
+        m = MultiIndex.from_arrays([[1, 3, 9],
+                                    ['C', 'D', 'E']],
+                                   names=['col3', 'col2'])
+        expected = DataFrame([3, 4, 5],
+                             index=m, columns=['col1'])
+
+        tm.assert_frame_equal(result, expected)
+
+        result = df.pivot_table(
+            'col1', index='col3', columns='col2', aggfunc=np.sum
+        )
+        expected = DataFrame([[3, np.NaN, np.NaN],
+                              [np.NaN, 4, np.NaN],
+                              [np.NaN, np.NaN, 5]],
+                             index=Index([1, 3, 9], name='col3'),
+                             columns=Index(['C', 'D', 'E'], name='col2'))
+
+        tm.assert_frame_equal(result, expected)
+
+        result = df.pivot_table('col1', index='col3', aggfunc=[np.sum])
+        m = MultiIndex.from_arrays([['sum'],
+                                    ['col1']])
+        expected = DataFrame([3, 4, 5],
+                             index=Index([1, 3, 9], name='col3'),
+                             columns=m)
+        tm.assert_frame_equal(result, expected)
+
 
 class TestCrosstab(tm.TestCase):