Apply condition on dataframe column pandas-CodePudding

If I have a dataframe like this :

date          A             B            C
01.01.2003    01.01.2003   
02.01.2003
03.01.2003                03.01.2003
05.01.2003    05.01.2003
06.01.2003                06.01.2003
08.01.2003    08.01.2003  08.01.2003    08.01.2003

And I want to change if value in column A, B, C are all equal I want to delete value in column A and B, leave the C. so the output

date          A             B            C
01.01.2003    01.01.2003   
02.01.2003
03.01.2003                03.01.2003
05.01.2003    05.01.2003
06.01.2003                06.01.2003
08.01.2003                             08.01.2003

I applied np.where but the error says condition does not apply on timestamp

np.where((df['A'] & df ['B'] == df['C]'),
           df['A'] & df['B], '')

thanks for the lead

CodePudding user response：

You can use pandas.DataFrame.loc with two conditions on row selection, namely A=B and B=C, and assign [None] to both A and B fields.

df.loc[(df['A']==df['B']) & (df['B']==df['C']), ['A', 'B']] = [[None, None]]

Output

         date            A           B           C
0  01.01.2003  01.,01.2003        None        None
1  02.01.2003         None        None        None
2  03.01.2003         None  03.01.2003        None
3  05.01.2003   05.01.2003        None        None
4  06.01.2003         None  06.01.2003        None
5  08.01.2003         None        None  08.01.2003

Check the demo here.

CodePudding user response：

Use boolean indexing with help of all:

df.loc[df[['A', 'B']].eq(df['C'], axis=0).all(axis=1), ['A', 'B']] = np.nan

Output:

         date           A           B           C
0  01.01.2003  01.01.2003        None        None
1  02.01.2003        None        None        None
2  03.01.2003         NaN  03.01.2003        None
3  05.01.2003  05.01.2003        None        None
4  06.01.2003         NaN  06.01.2003        None
5  08.01.2003         NaN         NaN  08.01.2003