使用pandas创建汇总表

Question

如何使用pandas从以下数据中获取汇总表：

ID  Condition   Confirmed
D0119   Bad Yes
D0119   Good    No
D0117   Bad Yes
D0110   Bad Undefined
D1011   Bad Yes
D1011   Good    Yes
D1001   Bad Yes
D1001   Bad Yes

必需的输出：

ID  Condition   Confirmed   %Bad
D0119   Bad,Good    Yes, No 50
D0117   Bad,Yes 100
D0110   Bad,Undefined   0
D1011   Bad,Good    Yes, Yes
D1001   Bad,Bad Yes, Yes    100

有人可以帮忙吗？感谢

Answer 1

你可以这样做：

In [123]: (df.assign(Bad=df.Condition=='Bad')
     ...:    .groupby('ID')
     ...:    .agg({'Condition':pd.Series.tolist,
     ...:          'Confirmed':pd.Series.tolist,
     ...:          'Bad':'mean'})
     ...: )
     ...:
Out[123]:
       Bad    Condition    Confirmed
ID
D0110  1.0        [Bad]  [Undefined]
D0117  1.0        [Bad]        [Yes]
D0119  0.5  [Bad, Good]    [Yes, No]
D1001  1.0   [Bad, Bad]   [Yes, Yes]
D1011  0.5  [Bad, Good]   [Yes, Yes]

垂直变体：

In [113]: df
Out[113]:
      ID Condition  Confirmed
0  D0119       Bad        Yes
1  D0119      Good         No
2  D0117       Bad        Yes
3  D0110       Bad  Undefined
4  D1011       Bad        Yes
5  D1011      Good        Yes
6  D1001       Bad        Yes
7  D1001       Bad        Yes

In [114]: g = df.assign(Bad=df.Condition=='Bad').groupby('ID')

In [115]: df['Bad'] = df['ID'].map((g.sum().div(g.size(), 0)*100).Bad)

In [116]: df
Out[116]:
      ID Condition  Confirmed    Bad
0  D0119       Bad        Yes   50.0
1  D0119      Good         No   50.0
2  D0117       Bad        Yes  100.0
3  D0110       Bad  Undefined  100.0
4  D1011       Bad        Yes   50.0
5  D1011      Good        Yes   50.0
6  D1001       Bad        Yes  100.0
7  D1001       Bad        Yes  100.0

Answer 2

考虑以下内容。

import pandas as pd

df = pd.DataFrame({'ID':['D0119', 'D0119', 'D0117', 'D0110', 'D1011', 'D1011', 'D1001', 'D1001'],
                   'Condition':['Bad', 'Good', 'Bad', 'Bad', 'Bad', 'Good', 'Bad', 'Bad'],
                   'Confirmed':['Yes', 'No', 'Yes', 'Undefined', 'Yes', 'Yes', 'Yes', 'Yes']})

df_grp = df.loc[df['Confirmed'] != 'Undefined'].groupby('ID')
summary = pd.DataFrame({'Condition':df_grp['Condition'],
                        'pnt_bad':df_grp['Condition'].apply(lambda x: sum(x=='Bad')/len(x))})

请注意，此方法不会保留仅具有“未定义”状态的记录的外观。