Question: Given DataFrame b, how can I replace the values of multiple columns, with one value, through boolean mask column identification?
What works, but I don't want:
b.iloc[:, 2:6] = "someConstantValue"
What doesn't work, but I want:
b.iloc[:, 2:6][b["columnA"] == "someCondition"]] = "someConstantValue"
Thanks for your time!
You can use iloc with Boolean indexing, but be careful. It works with Boolean arrays, not Boolean series. For example:
b.iloc[(b['A'] == 'a').values, 2:6] = 'someConstantValue'
As an aside, chained indexing is explicitly discouraged in the docs. There should never be a need to use chained indexing.
You have to use DataFrame.loc and for columns names by positions use indexing by b.columns[2:6].
b = pd.DataFrame({'A':list('abaaef'),
'B':[4,5,4,5,5,4],
'C':[7,8,9,4,2,3],
'D':[1,3,5,7,1,0],
'E':[5,3,6,9,2,4]})
print (b)
A B C D E
0 a 4 7 1 5
1 b 5 8 3 3
2 a 4 9 5 6
3 a 5 4 7 9
4 e 5 2 1 2
5 f 4 3 0 4
b.loc[b["A"] == "a", b.columns[2:6]] = 100
print (b)
A B C D E
0 a 4 100 100 100
1 b 5 8 3 3
2 a 4 100 100 100
3 a 5 100 100 100
4 e 5 2 1 2
5 f 4 3 0 4
