如何根据多个条件替换 2 个 dataframe 列中的值？

Question

I'm having columns in pandas dataframe which look like this:

Positive	Neutral	Negative
1	0	1
0	1	0

I want it to look like this:

Positive	Neutral	Negative	Mixed
0	0	0	1
0	1	0	0

First, I created column called mixed based on the fact that sentence is both positive and negative. Now, since I already have column "mixed", I do not need double information, so I would like to replace values in positive and negative columnwith 0 (only for mixed sentiment sentences). I've tried different variation of np.where but nothing seems to understand how to replace value in 2 columns based on condition from these 2 columns. Any suggestions? Thanks:)

Answer 1

It is a little bit unclear from your question. I guess your question (if i am correct) is to change from this:

Positive	Neutral	Negative
1	0	1

to this (since you have a new column called "mixed"):

Positive	Neutral	Negative	Mixed
0	0	0	1

If this is the case, then the code should be (i make the table with 3 rows instead of 1 for easy to see):

import pandas as pd

data = {'Positive': [1, 1, 1], 'Neutral': [0, 0, 0], 'Negative': [1, 1, 1]}
df = pd.DataFrame(data)
print(df)
print(\n)
x = (df['Positive'])
x[1] = 0 
print(df)

As the result, in column "Positive", the second row, value is changed from "1" to "0".

With the different index, you can adjust the code by yourself at x[i] = 0 . Similar code applied on "Negative" column.

Answer 2

You can just do it in two steps - set the mixed column - then set the pos/neg columns to 0.

>>> df['Mixed'] = 0
>>> df
   Positive  Neutral  Negative  Mixed
0         1        0         1      0
1         0        1         0      0
>>> rows = (df.Positive == 1) & (df.Negative == 1)
>>> df.loc[rows, 'Mixed'] = 1
>>> df.loc[rows, ['Positive', 'Negative']] = 0
>>> df
   Positive  Neutral  Negative  Mixed
0         0        0         0      1
1         0        1         0      0

You can use df.mask() if you want to do it all together.

>>> df
   Positive  Neutral  Negative  Mixed
0         1        0         1      0
1         0        1         0      0
>>> rows = (df.Positive == 1) & (df.Negative == 1)
>>> df.mask(rows, [0, 0, 0, 1])
   Positive  Neutral  Negative  Mixed
0         0        0         0      1
1         0        1         0      0