使用 .to_numpy() 将特定列从 Pandas Dataframe 的一行复制到另一行

Question

I have a Dataframe like this:

     UniqueID  CST  WEIGHT  VOLUME  PRODUCTIVITY
0  413-20012    3     123      12          1113
1  413-45365    1     889      75          6748
2  413-21165    8     554      13          4536
3  413-24354    1     387      35          7649
4  413-34658    2     121      88          2468
5  413-36889    4     105      76          3336
6  413-23457    5     355      42          7894
7  413-30089    5     146      10          9112
8  413-41158    5     453      91          4545
9  413-51015    9     654      66          2232

And I have a dictionary of parent:child mappings for the UniqueID's:

parent_child_dict = {
    '413-51015': '413-41158',
    '413-21165': '413-23457',
    '413-45365': '413-41158',
    '413-20012': '413-23457'
}

What I need to do is loop through the Dataframe, and replace the WEIGHT, VOLUME, and PRODUCTIVITY values of the 'child' UniqueID row with the values from the 'parent' UniqueID row, where resulting Dataframe would look like this:

     UniqueID  CST  WEIGHT  VOLUME  PRODUCTIVITY
0  413-20012    3     355      42          7894
1  413-45365    1     453      91          4545
2  413-21165    8     355      42          7894
3  413-24354    1     387      35          7649
4  413-34658    2     121      88          2468
5  413-36889    4     105      76          3336
6  413-23457    5     355      42          7894
7  413-30089    5     146      10          9112
8  413-41158    5     453      91          4545
9  413-51015    9     453      91          4545

I've experimented with several possible solutions, and the trouble I'm having is limiting the copy in such a way that the UniqueID and the CST values of the 'child' row are preserved, but the other values are copied over.

The closest I've gotten is a loop through the dictionary where each pairing gets fed into this:

df.loc[df['UniqueID'] == '413-51015'] = df.loc[df['UniqueID'] == '413-41158'].to_numpy()

This seems to nicely replace all values from one row to another.

Any help on the exceptions or a better solution overall would be extremely helpful. Thank you.

---

EDIT

I've looped the first solution into the columns that I want changed in the dataset like this:

columns = []
for col in df.columns:
    columns.append(col)
remove_perm = columns.remove('UniqueID')
remove_perm = columns.remove('CST')
print(columns)

OUTPUT

['WEIGHT', 'VOLUME', 'PRODUCTIVITY']

Then

for col in columns:
    s = df[['UniqueID', col]].set_index('UniqueID')
    df[col] = s.loc[df['UniqueID'].replace(parent_child_dict)].to_numpy()

This has resulted in the desired dataset.

Answer 1

replace and loc access:

s = df[['UniqueID','PRODUCTIVITY']].set_index('UniqueID')

# using to_numpy here :-)
df['PRODUCTIVITY'] = s.loc[df['UniqueID'].replace(parent_child_dict)].to_numpy()

Output:

    UniqueID  CST  WEIGHT  VOLUME  PRODUCTIVITY
0  413-20012    3     123      12          7894
1  413-45365    1     889      75          4545
2  413-21165    8     554      13          7894
3  413-24354    1     387      35          7649
4  413-34658    2     121      88          2468
5  413-36889    4     105      76          3336
6  413-23457    5     355      42          7894
7  413-30089    5     146      10          9112
8  413-41158    5     453      91          4545
9  413-51015    9     654      66          4545

Answer 2

First create a mapping out of your UniqueID and PRODUCTIVITY .

Then use your parent child to map your ids:

mapping = df.set_index('UniqueID')['PRODUCTIVITY'].to_dict()
df['PRODUCTIVITY'] = (
    df['UniqueID'].map(parent_child_dict).map(mapping).fillna(df['PRODUCTIVITY']).astype(int)
)
print(df)
    UniqueID  CST  WEIGHT  VOLUME  PRODUCTIVITY
0  413-20012    3     123      12          7894
1  413-45365    1     889      75          4545
2  413-21165    8     554      13          7894
3  413-24354    1     387      35          7649
4  413-34658    2     121      88          2468
5  413-36889    4     105      76          3336
6  413-23457    5     355      42          7894
7  413-30089    5     146      10          9112
8  413-41158    5     453      91          4545
9  413-51015    9     654      66          4545