Creating a new column with value dependent on on other columns values-CodePudding

Assuming I have a dataframe looking like below:

import pandas as pd
import numpy as np
d = {'Column 1': [10, 12,13,43,np.nan], 
    'Column2':[np.nan,7,np.nan,49,8]}
df = pd.DataFrame(d)

I would like to create a third column with a condition to take values from Column 2 unless they are NaNs. So looking like below:

I have found multiple topics/solutions where the condition was dependent on values in one column but could not find one where it had to provide data from more than one column.

CodePudding user response：

You could use mask:

df['Column3'] = df['Column2'].mask(df['Column2'].isna(), df['Column 1'])

A more generic version (uses any number of columns) would be to take the last valid value per row:

df['Column3'] = df.ffill(1).iloc[:,-1]

output:

   Column 1  Column2  Column3
0      10.0      NaN     10.0
1      12.0      7.0      7.0
2      13.0      NaN     13.0
3      43.0     49.0     49.0
4       NaN      8.0      8.0

CodePudding user response：

You could trying using np.where:

import pandas as pd
import numpy as np
d = {'Column 1': [10, 12,13,43,np.nan], 
    'Column2':[np.nan,7,np.nan,49,8]}
df = pd.DataFrame(d)

df['Column3'] = np.where(~pd.isna(df['Column2']), df['Column2'], df['Column 1'])
print(df)

   Column 1  Column2  Column3
0      10.0      NaN     10.0
1      12.0      7.0      7.0
2      13.0      NaN     13.0
3      43.0     49.0     49.0
4       NaN      8.0      8.0

CodePudding user response：

You only need:

df['Column3'] = df['Column2'].fillna(df['Column1'])

Or:

df['Column3'] = df['Column2'].combine_first(df['Column1'])