您可以使用ffill和bfill

>>> df1.replace("", pd.NA).ffill().bfill()
   CustomerNr Costs
0        1004     D
1        1004     D
2        1005     C
3        1010     A
4        1010     A
5        1010     A
6        1010     A
7        1004     A

你可以这样做：

import pandas as pd
import numpy as np

df1 = pd.DataFrame([[1004,''], [1004, 'D'],[1005, 'C'], 
                   [1010,'A'], [1010,''],[1010,''],[1010,''],[1004, '']], columns=['CustomerNr','Costs'])

df1['Costs'] = df1['Costs'].replace('', np.nan)  # Replace empty cells with NaNs
print(df1)
df1['Costs'] = df1.groupby('CustomerNr')['Costs'].fillna(method='ffill').fillna(method='bfill')

print(df1)

它给出了

CustomerNr Costs
0        1004     D
1        1004     D
2        1005     C
3        1010     A
4        1010     A
5        1010     A
6        1010     A
7        1004     D