基于现有列的下一行的元素创建新列

Question

我正在清理和重新构建数据框架。你知道吗

我有以下数据帧：

data= pd.DataFrame()
data['ID'] = [1,1,1,1,1,2,2,2,2,2]
data ['EventSecond'] = [1.5,2,2.5,3,3.8,4,4.8,6,7,8,]
data ['P1'] = ['A','B','C','D','E','F','A','D','E','G']
data ['Code'] = [12,13,16,9,9,0,4,13,14,16]
data ['status'] =['Pass','Pass','Pass','Pass','Pass','Pass','shot','shot','Pass','Pass']
data ['Accuracy']= ['Accurate','Accurate','Accurate','Accurate','Accurate','Not Accurate','Accurate','Accurate','Accurate','Not Accurate']

在这个数据帧中，我有ID，Eventsecond等。 我要做的是创建一个新列P2，其中包含列P1的下一行的元素，如果列accurity的元素是Accurate。有一点是，如果下面的ID列是不同的，我不会从下面的行中获取元素并将其留空 如果精度不准确，我将此行留空。你知道吗

问题补充

我只取status列的值为Pass的行。你知道吗

预期结果如下：

有人能提出建议吗？ 谢谢你

泽普。你知道吗

Answer 1

IIUC，你需要groupby和transform：

mask = (data['status'].isin(['Pass','pass']))
data.loc[mask,'P2'] = data[mask].groupby('ID')['P1'].transform(lambda x: x.shift(-1))
data.loc[data['Accuracy']=='Not Accurate','P2'] = np.nan

或仅使用过滤器：

mask = (data['status'].isin(['Pass','pass']))
data.loc[mask,'P2'] = data.loc[mask,'P1'].shift(-1)
mask2 = data['ID'].ne(data['ID'].shift(-1))|data['status'].eq('shot')|data['Accuracy'].eq('Not Accurate')
data.loc[mask2,'P2'] = ''

---

print(data)
   ID  EventSecond P1  Code status      Accuracy   P2
0   1          1.5  A    12   Pass      Accurate    B
1   1          2.0  B    13   Pass      Accurate    C
2   1          2.5  C    16   Pass      Accurate    D
3   1          3.0  D     9   Pass      Accurate    E
4   1          3.8  E     9   Pass      Accurate  NaN
5   2          4.0  F     0   Pass  Not Accurate  NaN
6   2          4.8  A     4   shot      Accurate  NaN
7   2          6.0  D    13   shot      Accurate  NaN
8   2          7.0  E    14   pass      Accurate    G
9   2          8.0  G    16   pass  Not Accurate  NaN

如果您选择的是空白，请使用：

print(data.fillna(''))

   ID  EventSecond P1  Code status      Accuracy P2
0   1          1.5  A    12   Pass      Accurate  B
1   1          2.0  B    13   Pass      Accurate  C
2   1          2.5  C    16   Pass      Accurate  D
3   1          3.0  D     9   Pass      Accurate  E
4   1          3.8  E     9   Pass      Accurate   
5   2          4.0  F     0   Pass  Not Accurate   
6   2          4.8  A     4   shot      Accurate   
7   2          6.0  D    13   shot      Accurate   
8   2          7.0  E    14   pass      Accurate  G
9   2          8.0  G    16   pass  Not Accurate   

Answer 2

因此，首先我将用P1中的shift创建P2，然后用您的条件创建一个mask，用loc空白来更改P2中的值，例如：

data['P2'] = data['P1'].shift(-1)
mask = ((data.Accuracy == 'Not Accurate') | 
        (data.status =='shot') | 
        (data.ID != data.ID.shift(-1)))
data.loc[mask,'P2'] = ''
print (data)
   ID  EventSecond P1  Code status      Accuracy P2
0   1          1.5  A    12   Pass      Accurate  B
1   1          2.0  B    13   Pass      Accurate  C
2   1          2.5  C    16   Pass      Accurate  D
3   1          3.0  D     9   Pass      Accurate  E
4   1          3.8  E     9   Pass      Accurate   
5   2          4.0  F     0   Pass  Not Accurate   
6   2          4.8  A     4   shot      Accurate   
7   2          6.0  D    13   shot      Accurate   
8   2          7.0  E    14   pass      Accurate  G
9   2          8.0  G    16   pass  Not Accurate   

编辑：你甚至可以像这样使用numpy.where

import numpy as np
data['P2'] = np.where(mask, '', data.P1.shift(-1))