Inplace operation on masked dataframe

Question

I'm a bit of noob with pandas and I'm trying to perform some calculations and modifications on parts of a masked dataframe using apply . The part I want to operate on is defined by my mask and I don't want to modify any non-masked values.

The thing is that I have no idea what is the proper way to put the result of the apply call on the masked dataframe back where it belongs in the original dataframe (or a copy of it, doesn't matter).

Here is a toy example of what I'm struggling with, I will try to make all values in the A column negative using a mask and apply:

import pandas as pd 
import numpy as np


def make_df():
    np.random.seed(4)
    df = pd.DataFrame(np.random.randn(5, 2),columns=["A","B"])
    return df

df = make_df()
mask = (df["A"]>0)

print(df)

          A         B
0  0.050562  0.499951
1 -0.995909  0.693599
2 -0.418302 -1.584577
3 -0.647707  0.598575
4  0.332250 -1.147477

The expected result is this :

          A         B
0 -0.050562  0.499951
1 -0.995909  0.693599
2 -0.418302 -1.584577
3 -0.647707  0.598575
4 -0.332250 -1.147477

What I hoped would work was this :

df = make_df()

df[mask]["A"] = df[mask]["A"].apply(lambda v: -v)
print(df)

          A         B
0  0.050562  0.499951
1 -0.995909  0.693599
2 -0.418302 -1.584577
3 -0.647707  0.598575
4  0.332250 -1.147477

But it fails with pandas warning me that df[mask]["A"] is a copy not a view so modifications on it do not affect df .

Answer 1

Try to use loc[] :

In [11]: df.loc[mask, 'A'] *= -1

In [12]: df
Out[12]:
          A         B
0 -0.050562  0.499951
1 -0.995909  0.693599
2 -0.418302 -1.584577
3 -0.647707  0.598575
4 -0.332250 -1.147477

Answer 2

你可以试试：

df.loc[df['A'] > 0,'A'] = -df.A