Create a column aggregating column names based on boolean state of a row in Python Pandas

I am attempting to scan a Pandas dataframe and identify all the columns that are True for a particular row and aggregate the column names in the output column.

Given a sample input as follows:

a       b       c
True    False   True
True    False   False

The goal is to accomplish the following output:

a       b       c     output
True    False   True  ['a', 'c']
True    False   False ['a']

My attempt here was to use np.where as exemplified below but this is rather inefficient to scale.

df['output_1'] = np.where(df['a']==True, 'a', '')
df['output_2'] = np.where(df['b']==True, 'b', '')
df['output_3'] = np.where(df['c']==True, 'c', '')
df['output'] = df['output_1'] + df['output_2'] + df['output_3']

>Solution :

You can use melt

Data:

data = {'a': {0: True, 1: True}, 'b': {0: False, 1: False}, 'c': {0: True, 1: False}}
df = pd.DataFrame(data=data)

Code:

df["output"] = (pd
                .melt(frame=df.reset_index(), id_vars=["index"])
                .query(expr="value.eq(True)")
                .groupby(by="index")
                .variable.apply(func=list)
                )

print(df)

Result:

      a      b      c  output
0  True  False   True  [a, c]
1  True  False  False     [a]