API: define _constructor_expanddim for subclassing Series and DataFrame

Author: sinhrks

pandas/core/frame.py

@@ -191,6 +191,11 @@ def _constructor(self):
 
     _constructor_sliced = Series
 
+    @property
+    def _constructor_expanddim(self):
+        from pandas.core.panel import Panel
+        return Panel
+
     def __init__(self, data=None, index=None, columns=None, dtype=None,
                  copy=False):
         if data is None:
@@ -1064,8 +1069,6 @@ def to_panel(self):
         -------
         panel : Panel
         """
-        from pandas.core.panel import Panel
-
         # only support this kind for now
         if (not isinstance(self.index, MultiIndex) or  # pragma: no cover
                 len(self.index.levels) != 2):
@@ -1103,7 +1106,7 @@ def to_panel(self):
                                               shape=shape,
                                               ref_items=selfsorted.columns)
 
-        return Panel(new_mgr)
+        return self._constructor_expanddim(new_mgr)
 
     to_wide = deprecate('to_wide', to_panel)
 
@@ -4413,12 +4416,12 @@ def mode(self, axis=0, numeric_only=False):
         """
         Gets the mode(s) of each element along the axis selected. Empty if nothing
         has 2+ occurrences. Adds a row for each mode per label, fills in gaps
-        with nan. 
-        
+        with nan.
+
         Note that there could be multiple values returned for the selected
-        axis (when more than one item share the maximum frequency), which is the 
-        reason why a dataframe is returned. If you want to impute missing values 
-        with the mode in a dataframe ``df``, you can just do this: 
+        axis (when more than one item share the maximum frequency), which is the
+        reason why a dataframe is returned. If you want to impute missing values
+        with the mode in a dataframe ``df``, you can just do this:
         ``df.fillna(df.mode().iloc[0])``
 
         Parameters

pandas/core/generic.py

@@ -154,6 +154,10 @@ def _local_dir(self):
     def _constructor_sliced(self):
         raise NotImplementedError
 
+    @property
+    def _constructor_expanddim(self):
+        raise NotImplementedError
+
     #----------------------------------------------------------------------
     # Axis
 

pandas/core/series.py

@@ -236,6 +236,11 @@ def from_array(cls, arr, index=None, name=None, dtype=None, copy=False,
     def _constructor(self):
         return Series
 
+    @property
+    def _constructor_expanddim(self):
+        from pandas.core.frame import DataFrame
+        return DataFrame
+
     # types
     @property
     def _can_hold_na(self):
@@ -1047,11 +1052,10 @@ def to_frame(self, name=None):
         -------
         data_frame : DataFrame
         """
-        from pandas.core.frame import DataFrame
         if name is None:
-            df = DataFrame(self)
+            df = self._constructor_expanddim(self)
         else:
-            df = DataFrame({name: self})
+            df = self._constructor_expanddim({name: self})
 
         return df
 

pandas/tests/test_frame.py

@@ -31,7 +31,7 @@
 import pandas.core.common as com
 import pandas.core.format as fmt
 import pandas.core.datetools as datetools
-from pandas import (DataFrame, Index, Series, notnull, isnull,
+from pandas import (DataFrame, Index, Series, Panel, notnull, isnull,
                     MultiIndex, DatetimeIndex, Timestamp, date_range,
                     read_csv, timedelta_range, Timedelta,
                     option_context)
@@ -14057,6 +14057,26 @@ def test_assign_bad(self):
         with tm.assertRaises(KeyError):
             df.assign(C=df.A, D=lambda x: x['A'] + x['C'])
 
+    def test_to_panel_expanddim(self):
+
+        class SubclassedFrame(DataFrame):
+            @property
+            def _constructor_expanddim(self):
+                return SubclassedPanel
+
+        class SubclassedPanel(Panel):
+            pass
+
+        index = MultiIndex.from_tuples([(0, 0), (0, 1), (0, 2)])
+        df = SubclassedFrame({'X':[1, 2, 3], 'Y': [4, 5, 6]}, index=index)
+        result = df.to_panel()
+        self.assertTrue(isinstance(result, SubclassedPanel))
+        expected = SubclassedPanel([[[1, 2, 3]], [[4, 5, 6]]],
+                                   items=['X', 'Y'], major_axis=[0],
+                                   minor_axis=[0, 1, 2])
+        tm.assert_panel_equal(result, expected)
+
+
 def skip_if_no_ne(engine='numexpr'):
     if engine == 'numexpr':
         try:

pandas/tests/test_series.py

@@ -6684,6 +6684,21 @@ def test_searchsorted_sorter(self):
         e = np.array([0, 2])
         tm.assert_array_equal(r, e)
 
+    def test_to_frame_expanddim(self):
+
+        class SubclassedSeries(Series):
+            @property
+            def _constructor_expanddim(self):
+                return SubclassedFrame
+
+        class SubclassedFrame(DataFrame):
+            pass
+
+        s = SubclassedSeries([1, 2, 3], name='X')
+        result = s.to_frame()
+        self.assertTrue(isinstance(result, SubclassedFrame))
+        expected = SubclassedFrame({'X': [1, 2, 3]})
+        assert_frame_equal(result, expected)
 
 
 class TestSeriesNonUnique(tm.TestCase):
@@ -6837,7 +6852,7 @@ def test_repeat(self):
     def test_unique_data_ownership(self):
         # it works! #1807
         Series(Series(["a", "c", "b"]).unique()).sort()
-    
+
     def test_datetime_timedelta_quantiles(self):
         # covers #9694
         self.assertTrue(pd.isnull(Series([],dtype='M8[ns]').quantile(.5)))

vb_suite/panel_methods.py

@@ -8,8 +8,8 @@
 # shift
 
 setup = common_setup + """
-index = date_range(start="2000", freq="D", periods=1000)
-panel = Panel(np.random.randn(100, len(index), 1000))
+index = date_range(start="2000", freq="D", periods=100)
+panel = Panel(np.random.randn(100, len(index), 100))
 """
 
 panel_shift = Benchmark('panel.shift(1)', setup,