Multi-hot-encode the target column data and inner join the resultant table with the original table.

add_multi_hot.py

#!/usr/bin/env python3
import re
from pathlib import Path

import pandas as pd
from sklearn.preprocessing import MultiLabelBinarizer


def add_multi_hot_to_df(df_in, target_col: str):
    """Multi-hot-encode the target column data
       and inner join the resultant table with the original table.
    
    Reference:
       https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MultiLabelBinarizer.html
    """
    # Safeguard for the MultiLabelBinarizer process
    df_in = df_in.fillna('void')
    # Convert data type of the target for MLB.
    pattern = re.compile(r'[\[\]\'\"]')
    # ['foo', 'bar'] :str -> 'foo, bar': str -> ['foo', 'bar']: list
    df_in[target_col] = df_in[target_col].apply(lambda x: pattern.sub('', x).split())

    mlb = MultiLabelBinarizer()
    mat = mlb.fit_transform(df_in[target_col])  # Multi-hot MATrix
    col = mlb.classes_  # Corresponds to the sorted set of classes
    dex = pd.DataFrame(mat, columns=col)  # Add a Header -> dexter table.
    # noinspection PyArgumentEqualDefault
    return pd.merge(df_in, dex, how='inner', left_index=True, right_index=True
                    ).replace('void', '')


if __name__ == '__main__':
    # Load the source.
    src_csv = Path(input('\033[93mSource CSV? >> \033[0m'))
    df = pd.read_csv(src_csv)
    target_col_ = input('\033[93mTarget column? >> \033[0m')
    # Main function call
    df = add_multi_hot_to_df(df, target_col_)
    df.to_csv(f'{src_csv.stem}_multi_hot.csv', index=False)
    print('\033[93mDONE!\033[0m')