根据其他列填充不同的列(第一列带有列标题，第二列带有要获取的值)

Question

我使用的是Python 3.6.9。

我被困在这样的数据帧上：

import pandas as pd
import numpy as np
dict_ = {'col1': [3.14, 28, -0.618, 1.159], 'col2': ['a_002_u', 'a_003_u', 'a_001_u', 'a_003_u'], 'a_001_u': [np.nan] * 4, 'a_002_u': [np.nan] * 4, 'a_003_u': [np.nan] * 4}
df = pd.DataFrame(dict_)
col1     col2  a_001_u  a_002_u  a_003_u
0   3.140  a_002_u      NaN      NaN      NaN
1  28.000  a_003_u      NaN      NaN      NaN
2  -0.618  a_001_u      NaN      NaN      NaN
3   1.159  a_003_u      NaN      NaN      NaN

我想得到这样的结果：

col1     col2  a_001_u  a_002_u  a_003_u
0   3.140  a_002_u      NaN     3.14      NaN
1  28.000  a_003_u      NaN      NaN   28.000
2  -0.618  a_001_u   -0.618      NaN      NaN
3   1.159  a_003_u      NaN      NaN    1.159

换句话说，我想根据"col2"中的列标题用"col1"值填充列"a_001_u"、"a_002_u"one_answers"a_003_u"。

这很容易解释，但我的印象是，它不太明显。有人想帮我吗？

Answer 1

您可以在使用set_index和unstack更改前2列的形状后使用fillna，如：

df = df.fillna(df.set_index('col2', append=True)['col1'].unstack())
print (df)
col1     col2  a_001_u  a_002_u  a_003_u
0   3.140  a_002_u      NaN     3.14      NaN
1  28.000  a_003_u      NaN      NaN   28.000
2  -0.618  a_001_u   -0.618      NaN      NaN
3   1.159  a_003_u      NaN      NaN    1.159

因为实际执行set_index和unstack确实会创建所需的其他列，而fillna将对缺少的值进行(行、列(填充

print(df.set_index('col2', append=True)['col1'].unstack())
col2  a_001_u  a_002_u  a_003_u
0         NaN     3.14      NaN
1         NaN      NaN   28.000
2      -0.618      NaN      NaN
3         NaN      NaN    1.159

注意：使用类似枢轴的df.pivot(columns='col2', values='col1')也可以获得相同的结果

Answer 2

您可以通过遍历行来编写它。

for index, row in df.iterrows():
row[row['col2']]=row['col1']

Answer 3

import pandas as pd
import numpy as np
dict_ = {'col1': [3.14, 28, -0.618, 1.159], 'col2': ['a_002_u', 'a_003_u', 'a_001_u', 'a_003_u'], 'a_001_u': [np.nan] * 4, 'a_002_u': [np.nan] * 4, 'a_003_u': [np.nan] * 4}
df = pd.DataFrame(dict_)
count = 0
for key in df['col2']:
df[key][count] = df['col1'][count]
count += 1
df