如果只有一列，为什么Pandas Transform失败 [英] Why Pandas Transform fails if you only have a single column
问题描述
After looking at this question I did some messing about and found this:
import pandas as pd
df = pd.DataFrame({'a':[1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4]})
df['num_totals'] = df.groupby('a').transform('count')
gives ValueError:
ValueError Traceback (most recent call last)
<ipython-input-38-157c6339ad93> in <module>()
3 #df = pd.DataFrame({'a':[1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4], 'b':[1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4]})
4 df = pd.DataFrame({'a':[1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4]})
----> 5 df['num_totals'] = df.groupby('a').transform('count')
6
7 #df['num_totals']=df.groupby('a')[['a']].transform('count')
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\frame.pyc in __setitem__(self, key, value)
2117 else:
2118 # set column
-> 2119 self._set_item(key, value)
2120
2121 def _setitem_slice(self, key, value):
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\frame.pyc in _set_item(self, key, value)
2164 """
2165 value = self._sanitize_column(key, value)
-> 2166 NDFrame._set_item(self, key, value)
2167
2168 def insert(self, loc, column, value, allow_duplicates=False):
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\generic.pyc in _set_item(self, key, value)
677
678 def _set_item(self, key, value):
--> 679 self._data.set(key, value)
680 self._clear_item_cache()
681
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\internals.pyc in set(self, item, value)
1779 except KeyError:
1780 # insert at end
-> 1781 self.insert(len(self.items), item, value)
1782
1783 self._known_consolidated = False
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\internals.pyc in insert(self, loc, item, value, allow_duplicates)
1793
1794 # new block
-> 1795 self._add_new_block(item, value, loc=loc)
1796
1797 except:
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\internals.pyc in _add_new_block(self, item, value, loc)
1909 loc = self.items.get_loc(item)
1910 new_block = make_block(value, self.items[loc:loc + 1].copy(),
-> 1911 self.items, fastpath=True)
1912 self.blocks.append(new_block)
1913
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\internals.pyc in make_block(values, items, ref_items, klass, fastpath, placement)
964 klass = ObjectBlock
965
--> 966 return klass(values, items, ref_items, ndim=values.ndim, fastpath=fastpath, placement=placement)
967
968 # TODO: flexible with index=None and/or items=None
C:\WinPython-64bit-2.7.5.3\python-2.7.5.amd64\lib\site-packages\pandas\core\internals.pyc in __init__(self, values, items, ref_items, ndim, fastpath, placement)
42 if len(items) != len(values):
43 raise ValueError('Wrong number of items passed %d, indices imply %d'
---> 44 % (len(items), len(values)))
45
46 self.set_ref_locs(placement)
ValueError: Wrong number of items passed 1, indices imply 0
但是，如果我有2列，则可以正常工作:
But if I have 2 columns then it works fine:
df = pd.DataFrame({'a':1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4],'b':1,1,1,1,2,2,3,3,3,4,4,4,4,4,4,4]})
df['num_totals'] = df.groupby('a').transform('count')
df
Out[40]:
a b num_totals
0 1 1 4
1 1 1 4
2 1 1 4
3 1 1 4
4 2 2 2
5 2 2 2
6 3 3 3
7 3 3 3
8 3 3 3
9 4 4 7
10 4 4 7
11 4 4 7
12 4 4 7
13 4 4 7
14 4 4 7
15 4 4 7
或者如果我使用单列df执行此操作:
or if I do this using a single column df:
df['num_totals']=df.groupby('a')[['a']].transform('count')
有类似的 SO帖子，但目前尚不清楚我为什么在上面的示例中，一系列操作会失败并且一个数据框应该工作，为什么具有两列或更多列会工作.
There is a similar SO post but it is unclear to me why a series should fail and a dataframe should work in the immediate above example, and why having 2 or more columns would work.
我正在使用Python 2.7 64位和Pandas 0.12
I am using Python 2.7 64-bit and Pandas 0.12
推荐答案
DF中的单列
如上所述，这将返回与原始尺寸相同的系列
Single Column in the DF
As you noted above, this returns a series the same size as the original
In [32]: df.groupby('a')['a'].transform('count')
Out[32]:
0 4
1 4
2 4
3 4
4 2
5 2
6 3
7 3
8 3
9 7
10 7
11 7
12 7
13 7
14 7
15 7
Name: a, dtype: int64
但是，这正在重现空白帧
However, this is returing an empty frame
In [33]: df.groupby('a').transform('count')
Out[33]:
Empty DataFrame
Columns: []
Index: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]
您不能将一个空框架作为一列分配给另一个框架，因为这本质上是一个模棱两可的分配(您可以证明它应该起作用")
you cannot assign a an empty frame as a column to another frame because this is essentially an ambiguous assignment (you can make a case that it should 'work' though)
两栏情况返回一个单列DataFrame
The two column case return a single-column DataFrame
In [42]: df2.groupby('a').transform('count')
Out[42]:
b
0 4
1 4
2 4
3 4
4 2
5 2
6 3
7 3
8 3
9 7
10 7
11 7
12 7
13 7
14 7
15 7
In [43]: type(df2.groupby('a').transform('count'))
Out[43]: pandas.core.frame.DataFrame
Or a series
In [45]: df2.groupby('a')['a'].transform('count')
Out[45]:
0 4
1 4
2 4
3 4
4 2
5 2
6 3
7 3
8 3
9 7
10 7
11 7
12 7
13 7
14 7
15 7
Name: a, dtype: int64
In [46]: type(df.groupby('a')['a'].transform('count'))
Out[46]: pandas.core.series.Series
之所以行之有效"，是因为熊猫确实允许分配单个列框架，因为它将采用基础系列.
This 'works' because pandas DOES allow assignment of a single column frame to work, as it will take the underlying series.
因此，熊猫实际上正在尝试提供帮助.就是说，我发现这是一条不清楚的错误消息，用于尝试分配一个空框架.
So pandas is actually trying to be helpful. That said, I find this an unclear error message for trying to assign an empty frame.
这篇关于如果只有一列，为什么Pandas Transform失败的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
