Map function modifies more than one column

January 12, 2023

I trying to modify the value of just a single columns in a pandas dataframe, however, when I apply the map() function, another unrelated column also changes its values. How can I modify just a single column?

columns = [[[776, 114, 67, 47], [776, 167, 67, 52], [776, 224, 67, 62],[776, 291, 67, 58],[776, 355, 67, 52]], 
    [[849, 114, 76, 48], [849, 167, 76, 52],[849, 225, 76, 61],[849, 291, 76, 59],[849, 355, 75, 53]],
    [[930, 115, 113, 47], [930, 168, 113, 51], [930, 225, 113, 61], [930, 292, 113, 58], [930, 355, 113, 53]]]

dic = {'A': columns[0], 'B': columns[0], 'C': columns[1], 'D': columns[2]}
df = pd.DataFrame.from_dict(dic)


def subs(x):
    x[0] = 230
    x[2] = 533
    return x


df['A'] = df['A'].map(subs)

                 A                   B                  C                    D
0   [230, 114, 533, 47] [230, 114, 533, 47] [849, 114, 76, 48]  [930, 115, 113, 47]
1   [230, 167, 533, 52] [230, 167, 533, 52] [849, 167, 76, 52]  [930, 168, 113, 51]
2   [230, 224, 533, 62] [230, 224, 533, 62] [849, 225, 76, 61]  [930, 225, 113, 61]
3   [230, 291, 533, 58] [230, 291, 533, 58] [849, 291, 76, 59]  [930, 292, 113, 58]
4   [230, 355, 533, 52] [230, 355, 533, 52] [849, 355, 75, 53]  [930, 355, 113, 53]

I need it to return as follows:

                 A                   B                  C                    D
0   [230, 114, 533, 47] [776, 114, 67, 47]  [849, 114, 76, 48]  [930, 115, 113, 47]
1   [230, 167, 533, 52] [776, 167, 67, 52]  [849, 167, 76, 52]  [930, 168, 113, 51]
2   [230, 224, 533, 62] [776, 224, 67, 62]  [849, 225, 76, 61]  [930, 225, 113, 61]
3   [230, 291, 533, 58] [776, 291, 67, 58]  [849, 291, 76, 59]  [930, 292, 113, 58]
4   [230, 355, 533, 52] [776, 355, 67, 52]  [849, 355, 75, 53]  [930, 355, 113, 53]

>Solution :

A and B use the same underlying lists.

Use deepcopy(columns[0]) to make it independent:

from copy import deepcopy

dic = {'A': columns[0], 'B': deepcopy(columns[0]), 'C': columns[1], 'D': columns[2]}
df = pd.DataFrame.from_dict(dic)

Note that this still modifies columns, if you want to avoid this use deepcopy for all columns.

Output:

                     A                   B                   C                    D
0  [230, 114, 533, 47]  [776, 114, 67, 47]  [849, 114, 76, 48]  [930, 115, 113, 47]
1  [230, 167, 533, 52]  [776, 167, 67, 52]  [849, 167, 76, 52]  [930, 168, 113, 51]
2  [230, 224, 533, 62]  [776, 224, 67, 62]  [849, 225, 76, 61]  [930, 225, 113, 61]
3  [230, 291, 533, 58]  [776, 291, 67, 58]  [849, 291, 76, 59]  [930, 292, 113, 58]
4  [230, 355, 533, 52]  [776, 355, 67, 52]  [849, 355, 75, 53]  [930, 355, 113, 53]