BUG: GH11847 Unstack with mixed dtypes coerces everything to object

closes #11847

Changed the way
in which the original data frame is copied (dropped use of .values,
since it does not preserve dtypes).

Author: Pawel Kordek <[email protected]>

Closes #14053 from kordek/#11847 and squashes the following commits:

6a381ce [Pawel Kordek] BUG: GH11847 Unstack with mixed dtypes coerces everything to object

Author: kordek

doc/source/whatsnew/v0.19.2.txt

@@ -76,3 +76,5 @@ Bug Fixes
 
 
 - Explicit check in ``to_stata`` and ``StataWriter`` for out-of-range values when writing doubles (:issue:`14618`)
+
+- Bug in ``unstack()`` if called with a list of column(s) as an argument, regardless of the dtypes of all columns, they get coerced to ``object`` (:issue:`11847`)

pandas/core/reshape.py

@@ -277,7 +277,8 @@ def _unstack_multiple(data, clocs):
                              verify_integrity=False)
 
     if isinstance(data, Series):
-        dummy = Series(data.values, index=dummy_index)
+        dummy = data.copy()
+        dummy.index = dummy_index
         unstacked = dummy.unstack('__placeholder__')
         new_levels = clevels
         new_names = cnames
@@ -292,7 +293,8 @@ def _unstack_multiple(data, clocs):
 
             return result
 
-        dummy = DataFrame(data.values, index=dummy_index, columns=data.columns)
+        dummy = data.copy()
+        dummy.index = dummy_index
 
         unstacked = dummy.unstack('__placeholder__')
         if isinstance(unstacked, Series):

pandas/tests/frame/test_reshape.py

@@ -282,6 +282,46 @@ def test_unstack_fill_frame_categorical(self):
                              index=list('xyz'))
         assert_frame_equal(result, expected)
 
+    def test_unstack_preserve_dtypes(self):
+        # Checks fix for #11847
+        df = pd.DataFrame(dict(state=['IL', 'MI', 'NC'],
+                               index=['a', 'b', 'c'],
+                               some_categories=pd.Series(['a', 'b', 'c']
+                                                         ).astype('category'),
+                               A=np.random.rand(3),
+                               B=1,
+                               C='foo',
+                               D=pd.Timestamp('20010102'),
+                               E=pd.Series([1.0, 50.0, 100.0]
+                                           ).astype('float32'),
+                               F=pd.Series([3.0, 4.0, 5.0]).astype('float64'),
+                               G=False,
+                               H=pd.Series([1, 200, 923442], dtype='int8')))
+
+        def unstack_and_compare(df, column_name):
+            unstacked1 = df.unstack([column_name])
+            unstacked2 = df.unstack(column_name)
+            assert_frame_equal(unstacked1, unstacked2)
+
+        df1 = df.set_index(['state', 'index'])
+        unstack_and_compare(df1, 'index')
+
+        df1 = df.set_index(['state', 'some_categories'])
+        unstack_and_compare(df1, 'some_categories')
+
+        df1 = df.set_index(['F', 'C'])
+        unstack_and_compare(df1, 'F')
+
+        df1 = df.set_index(['G', 'B', 'state'])
+        unstack_and_compare(df1, 'B')
+
+        df1 = df.set_index(['E', 'A'])
+        unstack_and_compare(df1, 'E')
+
+        df1 = df.set_index(['state', 'index'])
+        s = df1['A']
+        unstack_and_compare(s, 'index')
+
     def test_stack_ints(self):
         columns = MultiIndex.from_tuples(list(itertools.product(range(3),
                                                                 repeat=3)))