根据多个条件将一列拆分为几列并分组

Posted 2023-03-11

【中文标题】根据多个条件将一列拆分为几列并分组

【英文标题】：Split a column into several columns based on several conditions and group by

【发布时间】：2021-11-25 23:49:18

【问题描述】：

我有一个示例数据框，如下所示。

import pandas as pd

data = 'ID':['A','A','A','A','A','A','A','A','A','C','C','C','C','C','C','C','C'],
    'Week': ['Week1','Week1','Week1','Week1','Week2','Week2','Week2','Week2','Week3',
             'Week1','Week1','Week1','Week1','Week2','Week2','Week2','Week2'],
    'Risk':['High','','','','','','','','','High','','','','','','',''],
    'Testing':['','Pos','','Neg','','','','','Pos', '', '','','Neg','','','','Pos'],
    'Week1_adher':['','','','','','','','','', '','','','','','','',''],
    'Week2_adher':['','','','','','','','','','','','','','','','',''],
    'Week3_adher':['','','','','','','','','','','','','','','','','']
    
df1 = pd.DataFrame(data)
df1 

现在我想计算每个参与者每周的依从性。其计算如下： 如果参与者在一周内的测试栏中有 2 个或更多条目（正面/负面），则该周的坚持为“是”，否则为“否”

例如，对于参与者 A，第 1 周_adherence 为“是”，因为它在第 1 周的测试列中有 2 个条目。 Week2_adherence 为“否”

并且我希望将整周的依从性结果显示在每个参与者的第一行。

最终的数据框应该如下图所示。

我已经坚持了很长一段时间了。任何帮助是极大的赞赏。谢谢。

【参考方案1】：

试试：

adher = (df1.Testing.ne('')       # check for non-empty string
    .groupby([df1.ID, df1.Week])  # groupby ID and week
    .sum().ge(2)                  # count and check >= 2
    .unstack(fill_value=False)
    .replace(True:'Yes', False:'No')
    .add_suffix('_adher')
)

# the first lines
mask = ~df1['ID'].duplicated()

df1.loc[mask, adher.columns] = adher.loc[df1.loc[mask,'ID']].values

输出：

   ID   Week  Risk   Testing Week1_adher Week2_adher Week3_adher
0   A  Week1  High                   Yes          No          No
1   A  Week1             Pos                                    
2   A  Week1                                                    
3   A  Week1             Neg                                    
4   A  Week2                                                    
5   A  Week2                                                    
6   A  Week2                                                    
7   A  Week2                                                    
8   A  Week3             Pos                                    
9   C  Week1  High                    No          No          No
10  C  Week1                                                    
11  C  Week1                                                    
12  C  Week1        Negative                                    
13  C  Week2                                                    
14  C  Week2                                                    
15  C  Week2                                                    
16  C  Week2        Positive                                    

【讨论】：

适用于非空字符串（第一行）。如果它是 np.nan 值而不是空字符串怎么办？只需用 testing.ne(np.nan) 替换它？我试过这个，但不起作用。

使用notna()而不是ne()来检查非nan值。

知道了。非常感谢！。解决方案也很优雅。