ENH: Declare a BoolBlock as a NumericBlock

BUG: GH2641 fixes "df.decribe() with boolean column"

This change will make all numeric operations on boolean data work, by
just transparently treating them as integers values 1 and 0.  This is
not pandas specific behavior, this is the default operations of both
numpy and python.

RELEASE.rst

@@ -124,6 +124,8 @@ pandas 0.11.0
     knows how many columns to expect in the result) (GH2981_)
   - On a mixed DataFrame, allow setting with indexers with ndarray/DataFrame
     on rhs (GH3216_)
+  - Treat boolean values as integers (values 1 and 0) for numeric
+    operations. (GH2641_)
 
 **API Changes**
 
@@ -350,6 +352,7 @@ pandas 0.11.0
 .. _GH2747: https://github.com/pydata/pandas/issues/2747
 .. _GH2816: https://github.com/pydata/pandas/issues/2816
 .. _GH3216: https://github.com/pydata/pandas/issues/3216
+.. _GH2641: https://github.com/pydata/pandas/issues/2641
 
 pandas 0.10.1
 =============

doc/source/v0.11.0.txt

@@ -304,6 +304,9 @@ Enhancements
   - added option `display.with_wmp_style` providing a sleeker visual style
     for plots. Based on https://gist.github.com/huyng/816622 (GH3075_).
 
+  - Treat boolean values as integers (values 1 and 0) for numeric
+    operations. (GH2641_)
+
 See the `full release notes
 <https://github.com/pydata/pandas/blob/master/RELEASE.rst>`__ or issue tracker
 on GitHub for a complete list.
@@ -328,3 +331,4 @@ on GitHub for a complete list.
 .. _GH3059: https://github.com/pydata/pandas/issues/3059
 .. _GH3070: https://github.com/pydata/pandas/issues/3070
 .. _GH3075: https://github.com/pydata/pandas/issues/3075
+.. _GH2641: https://github.com/pydata/pandas/issues/2641

pandas/core/internals.py

@@ -628,7 +628,7 @@ def should_store(self, value):
         return com.is_integer_dtype(value) and value.dtype == self.dtype
 
 
-class BoolBlock(Block):
+class BoolBlock(NumericBlock):
     is_bool = True
     _can_hold_na = False
 
@@ -641,9 +641,6 @@ def _try_cast(self, element):
         except:  # pragma: no cover
             return element
 
-    def _try_cast_result(self, result):
-        return _possibly_downcast_to_dtype(result, self.dtype)
-
     def should_store(self, value):
         return issubclass(value.dtype.type, np.bool_)
 

pandas/tests/test_frame.py

@@ -7899,15 +7899,14 @@ def test_dataframe_clip(self):
     def test_get_X_columns(self):
         # numeric and object columns
 
-        # Booleans get casted to float in DataFrame, so skip for now
         df = DataFrame({'a': [1, 2, 3],
-                        # 'b' : [True, False, True],
+                        'b' : [True, False, True],
                         'c': ['foo', 'bar', 'baz'],
                         'd': [None, None, None],
                         'e': [3.14, 0.577, 2.773]})
 
         self.assert_(np.array_equal(df._get_numeric_data().columns,
-                                    ['a', 'e']))
+                                    ['a', 'b', 'e']))
 
     def test_get_numeric_data(self):
         intname = np.dtype(np.int_).name
@@ -7939,6 +7938,30 @@ def test_get_numeric_data(self):
         expected = df.ix[:, []]
         assert_frame_equal(result, expected)
 
+    def test_bool_describe_in_mixed_frame(self):
+        df = DataFrame({
+            'string_data': ['a', 'b', 'c', 'd', 'e'],
+            'bool_data': [True, True, False, False, False],
+            'int_data': [10, 20, 30, 40, 50],
+        })
+
+        # Boolean data and integer data is included in .describe() output, string data isn't
+        self.assert_(np.array_equal(df.describe().columns, ['bool_data', 'int_data']))
+
+        bool_describe = df.describe()['bool_data']
+
+        # Both the min and the max values should stay booleans
+        self.assert_(bool_describe['min'].dtype == np.bool_)
+        self.assert_(bool_describe['max'].dtype == np.bool_)
+
+        self.assert_(bool_describe['min'] == False)
+        self.assert_(bool_describe['max'] == True)
+
+        # For numeric operations, like mean or median, the values True/False are cast to
+        # the integer values 1 and 0
+        assert_almost_equal(bool_describe['mean'], 0.4)
+        assert_almost_equal(bool_describe['50%'], 0)
+
     def test_count(self):
         f = lambda s: notnull(s).sum()
         self._check_stat_op('count', f,

pandas/tests/test_internals.py

@@ -497,7 +497,7 @@ def test_get_numeric_data(self):
                         'bool': bool_ser, 'obj': obj_ser,
                         'dt': dt_ser})
         xp = DataFrame({'int': int_ser, 'float': float_ser,
-                        'complex': complex_ser})
+                        'complex': complex_ser, 'bool': bool_ser})
         rs = DataFrame(df._data.get_numeric_data())
         assert_frame_equal(xp, rs)