python - Pandas hub_table 与 aggfunc 在不同数据上的工作方式有所不同

Question

我有一个包含日期、买入和卖出值的数据框，并在该数据框上执行pivot_table获取每个有效日期的所有买入/卖出值(value)。

data = [('20170325', 'Buy', 400 ),
        ('20170325', 'Buy', 401 ),
        ('20170323', 'Buy', 400 ),
        ('20170324', 'Sell', 400 )]

testDf = pd.DataFrame(data, columns=['Date', 'Scenario', 'Value'])
df1 = pd.pivot_table(testDf, columns='Scenario', index='Date', values='Value', 
                     fill_value = '', aggfunc=lambda x: x.tolist() if len(x)>1 else x)
df1
#    Scenario         Buy Sell
#    Date                     
#    20170323         400     
#    20170324              400
#    20170325  [400, 401]

但是当我的数据发生一点变化时，pivot_table 会抛出错误:ValueError:函数不会减少。无法理解为什么。这是不同数据的错误。请注意，日期的20170325已更改为20170321。

data = [('20170321', 'Buy', 400 ),
        ('20170321', 'Buy', 401 ),
        ('20170323', 'Buy', 400 ),
        ('20170324', 'Sell', 400 )]

testDf = pd.DataFrame(data, columns=['Date', 'Scenario', 'Value'])
df1 = pd.pivot_table(testDf, columns='Scenario', index='Date', values='Value', 
                     fill_value = '',aggfunc=lambda x: x.tolist() if len(x)>1 else x)

Traceback (most recent call last): File "", line 1, in
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\tools\pivot.py", line 114, in pivot_table
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\core\groupby.py", line 729, in agg
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\core\groupby.py", line 2978, in aggregate
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\core\groupby.py", line 1227, in _python_agg_general
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\core\groupby.py", line 1733, in agg_series
File "ext2\vc12_win32\lib\python2.7\site-packages\pandas\core\groupby.py", line 1767, in _aggregate_series_pure_python
ValueError: Function does not reduce

Answer 1

甚至考虑使用 pivot 的 groupby:

df1 = testDf.groupby(['Date', 'Scenario'])['Value'].apply(lambda x: list(x)).reset_index()\
            .pivot(index='Date', columns='Scenario', values='Value').fillna('')

# Scenario         Buy   Sell
# Date                       
# 20170323       [400]       
# 20170324              [400]
# 20170325  [400, 401]       

# Scenario         Buy   Sell
# Date                       
# 20170321  [400, 401]       
# 20170323       [400]       
# 20170324              [400]