python - 使用 DataFrame.sort_index(axis=1) 时出现意外顺序。第一列最后列出

Question

我有一个像这样填充的示例 DataFrame:

               Alpha      Beta     Gamma     Delta   Epsilon
Date
2017-01-02  0.854046       NaN  0.681606  0.883779  0.680304
2017-01-01  0.573784  0.407917  0.446668  0.463504  0.136830
2017-01-03  0.556100  0.849009  0.389748       NaN  0.777201

当我按 axis=1(列标题)排序时，“Alpha”列被放在最后一个位置:

df_sorted = df1.sort_index(axis=1)
print df_sorted

输出:

                Beta     Delta   Epsilon     Gamma     Alpha
Date
2017-01-02       NaN  0.883779  0.680304  0.681606  0.854046
2017-01-01  0.407917  0.463504  0.136830  0.446668  0.573784
2017-01-03  0.849009       NaN  0.777201  0.389748  0.556100

谁能解释列的字母数字排序？

谢谢!

Answer 1

也许你在一些列名前面有空格，使用 @rahlf23 设置并进行一些修改:

df = pd.DataFrame([['2017-01-02',  ' 0.854046',       np.nan,  '0.681606',  '0.883779',  '0.680304'],
                    ['2017-01-01',  '0.573784',  '0.407917',  '0.446668',  '0.463504',  '0.136830'],
                    ['2017-01-03',  '0.556100',  '0.849009',  '0.389748',       np.nan,  '0.777201']],
                    columns=['Date', ' Beta',      ' Gamma',     ' Delta',     'Alpha',   ' Epsilon']).set_index('Date')

df.sort_index(axis=1)

输出:

                 Beta     Delta   Epsilon     Gamma     Alpha
Date                                                         
2017-01-02   0.854046  0.681606  0.680304       NaN  0.883779
2017-01-01   0.573784  0.446668  0.136830  0.407917  0.463504
2017-01-03   0.556100  0.389748  0.777201  0.849009       NaN

要识别它，您可以使用 df.to_dict():

{' Beta': {'2017-01-01': '0.573784',
  '2017-01-02': ' 0.854046',
  '2017-01-03': '0.556100'},
 ' Delta': {'2017-01-01': '0.446668',
  '2017-01-02': '0.681606',
  '2017-01-03': '0.389748'},
 ' Epsilon': {'2017-01-01': '0.136830',
  '2017-01-02': '0.680304',
  '2017-01-03': '0.777201'},
 ' Gamma': {'2017-01-01': '0.407917',
  '2017-01-02': nan,
  '2017-01-03': '0.849009'},
 'Alpha': {'2017-01-01': '0.463504',
  '2017-01-02': '0.883779',
  '2017-01-03': nan}}

要修复，您可以使用 .str.strip()

df.columns = df.columns.str.strip()
df.sort_index(axis=1)

输出:

               Alpha       Beta     Delta   Epsilon     Gamma
Date                                                         
2017-01-02  0.883779   0.854046  0.681606  0.680304       NaN
2017-01-01  0.463504   0.573784  0.446668  0.136830  0.407917
2017-01-03       NaN   0.556100  0.389748  0.777201  0.849009