Converting binary encoding to classes multilabel python

Question

I have data like this in dataframe

Text	Label1	Label2	label3	label4	Labels
Hello my name is john	1	0	1	0

and I want to fill the labels column based on the ones and zeros to be like this

Text	Label1	Label2	label3	label4	Labels
Hello my name is john	1	0	1	0	['Label1','Label3']

How can I do in python?

I was thinking ['Label1', 'Label**2**'] was a typo that should be ['Label1', 'Label**3**'], can you confirm or not? — Joffan, Apr 22 '21 at 16:09

score 1 · Accepted Answer · answered Apr 22 '21 at 17:13

Ok, this is an untested suggestion from this answer. Try gathering the appropriate column labels then using DataFrame.apply on a suitable function:

test_cols = [c for c in df.columns if c[:5].lower() == 'label']
test_cols.remove('Labels')

def aggLabels(aSeries):
    return [lab for lab in test_cols if aSeries[lab]==1]

df['Labels'] = df.apply(aggLabels, axis=1)

As I say this is untested; there may be code tweaks required.

Converting binary encoding to classes multilabel python

1 Answers1