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']