Question

我想在一列中显示我拥有的组数，然后只显示那些具有超过特定数字的组。

考虑这个例子：

import pandas as pd

df = pd.DataFrame(
    {
        'ColA': 'A A A B B C C C C D E E F F F F F F F G G H'.split(),
        'ColB': '1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2'.split()
    }
)

如果我这样分组：

print df.groupby(['ColA']).agg(['count'])

我得到这样的输出：

现在，如果我只想在计数超过2时显示上面的行，我该怎么做？我希望不包含B，D，E，G或H

的输出

我尝试了以下两行，并且都返回了相同的错误：

print df.loc[df.groupby(['ColA']).agg(['count']) > 2]
print df.loc[df.groupby(['ColA']).agg(['count'])['ColB'] > 2]

错误：

Traceback (most recent call last):
  File "C:/scratches/scratch_3", line 11, in <module>
    print df.loc[df.groupby(['ColA']).agg(['count'])['ColB'] > 2]
  File "C:\Anaconda\lib\site-packages\pandas\core\indexing.py", line 1189, in __getitem__
    return self._getitem_axis(key, axis=0)
  File "C:\Anaconda\lib\site-packages\pandas\core\indexing.py", line 1321, in _getitem_axis
    raise ValueError('Cannot index with multidimensional key')
ValueError: Cannot index with multidimensional key

PabTorre提供的answer似乎不适用于较新版本的pandas。我正在使用0.16.2当我使用该答案时，我在此行收到以下错误：

print df_count[df_count.values>2]

Traceback (most recent call last):
  File "C:/scratches/scratch_3", line 10, in <module>
    print df_count[df_count.values>2]
  File "C:\Anaconda\lib\site-packages\pandas\core\frame.py", line 1791, in __getitem__
    return self._getitem_array(key)
  File "C:\Anaconda\lib\site-packages\pandas\core\frame.py", line 1833, in _getitem_array
    return self.take(indexer, axis=0, convert=False)
  File "C:\Anaconda\lib\site-packages\pandas\core\generic.py", line 1358, in take
    convert=True, verify=True)
  File "C:\Anaconda\lib\site-packages\pandas\core\internals.py", line 3275, in take
    axis=axis, allow_dups=True)
  File "C:\Anaconda\lib\site-packages\pandas\core\internals.py", line 3162, in reindex_indexer
    for blk in self.blocks]
  File "C:\Anaconda\lib\site-packages\pandas\core\internals.py", line 857, in take_nd
    allow_fill=True, fill_value=fill_value)
  File "C:\Anaconda\lib\site-packages\pandas\core\common.py", line 844, in take_nd
    func(arr, indexer, out, fill_value)
  File "pandas\src\generated.pyx", line 5779, in pandas.algos.take_2d_axis1_object_object (pandas\algos.c:107426)
  File "stringsource", line 614, in View.MemoryView.memoryview_cwrapper (pandas\algos.c:187433)
  File "stringsource", line 321, in View.MemoryView.memoryview.__cinit__ (pandas\algos.c:184022)
ValueError: buffer source array is read-only

Answer 1

您上次查询的问题：

print df.loc[df.groupby(['ColA']).agg(['count']) > 2]

df.loc []是否需要一系列22个布尔对象。相反，它获得了一系列8个对象：

>>> df.groupby(['ColA']).agg(['count']) > 2
       ColB
      count
ColA       
A      True
B     False
C      True
D     False
E     False
F      True
G     False
H     False

因此它不知道如何将它们组合在一起。

但是有一个解决方案。 :)

首先，让我们将聚合df分配给一个新对象。

>>> df_count = df.groupby(['ColA']).agg(['count']).ColB
>>> df_count.columns=['ColB']

然后我们可以轻松过滤它

>>> df_count[df_count.ColB.values>2]
       ColB

ColA       
A         3
C         4
F         7

然后我们可以使用过滤后的df返回并过滤原始df

>>> df_filtered=df_count[df_count.ColB.values>2]
>>> df[df.ColA.isin(df_filtered.index)]
   ColA ColB
0     A    1
1     A    2
2     A    3
5     C    6
6     C    7
7     C    8
8     C    9
12    F    3
13    F    4
14    F    5
15    F    6
16    F    7
17    F    8
18    F    9

Pandas数据帧仅由具有高于X的计数器的行组成

1 个答案: