CODEWITHSUNDEEP
pythonscikit-learnfeature-extraction
Mukesh Gupta
Mukesh Gupta

Reputation: 1433

Preprocess multivalues feature?

How to preprocess multivalued feature (genre_ids)?

As I need to assign one numeric value to each row of one column, but how efficiently I can preprocess this genre_ids?

msno  |city |bd  |gender| song_id| song_length| genre_ids| artist_name 
1      12   33    0       11       222          372                89   
2      11   23    1       1        202          372|374|375        99
3      14   22    1       2        300          386                78

Upvotes: 0

Views: 48

Answers (1)

Eduard Ilyasov
Eduard Ilyasov

Reputation: 3308

You can use pandas package to solve your task:

import pandas as pd

df = pd.DataFrame({'msno': [1, 2, 3],
                   'city': [12, 11, 14],
                   'bd': [33, 23, 22],
                   'gender': [0, 1, 1],
                   'song_id': [11, 1, 2],
                   'song_length': [222, 202, 300],
                   'genre_ids': ['372', '372|374|375', '386'],
                   'artist_name': [89, 99, 78]})

genre_ids_dummies = (pd.get_dummies(df.genre_ids.apply(lambda x: x.split('|'))
                                                .apply(pd.Series)
                                                .stack()).sum(level=0))
df_processed = pd.concat(objs=[df.drop(labels='genre_ids', axis=1), 
                               genre_ids_dummies], axis=1)

Upvotes: 1

Related Questions