两个值匹配pandas时的累计计数

Posted 2021-05-04

我正在尝试创建一个新的Column，在单独的cumulative count中显示基于值的columns。

所以对于下面的代码，我正在尝试根据Cause和Answer Columns创建两个新列。因此，对于Column Answer中的值，如果In位于Column Cause，我想在新列中提供累积计数。

import pandas as pd

d = ({
    'Cause' : ['In','','','In','','In','In'],
    'Answer' : ['Yes','No','Maybe','No','Yes','No','Yes'],
    })

df = pd.DataFrame(d)

输出：

  Answer Cause
0    Yes    In
1     No      
2  Maybe      
3     No    In
4    Yes      
5     No    In
6    Yes    In

预期产出：

  Answer Cause Count_No Count_Yes
0    Yes    In                  1
1     No                         
2  Maybe                         
3     No    In        1          
4    Yes                         
5     No    In        2          
6    Yes    In                  2

我尝试过以下但是出错了。

df['cumsum'] = df.groupby(['Answer'])['Cause'].cumsum()

答案

没有for循环： - ）

s=df.loc[df.Cause=='In'].Answer.str.get_dummies()
pd.concat([df,s.cumsum().mask(s!=1,'')],axis=1).fillna('')
Out[62]: 
  Answer Cause No Yes
0    Yes    In      1
1     No             
2  Maybe             
3     No    In  1    
4    Yes             
5     No    In  2    
6    Yes    In      2

另一答案

这是一种方式 -

for val in ['Yes', 'No']:
    cond = df.Answer.eq(val) & df.Cause.eq('In')
    df.loc[cond, 'Count_' + val] = cond[cond].cumsum()

df
#  Cause Answer  Count_Yes  Count_No
#0    In    Yes        1.0       NaN
#1           No        NaN       NaN
#2        Maybe        NaN       NaN
#3    In     No        NaN       1.0
#4          Yes        NaN       NaN
#5    In     No        NaN       2.0
#6    In    Yes        2.0       NaN