Muzyka i data science

import spotipy
from spotipy import util

import pandas as pd

SpotifyClientID='xxxxxxxxxxxxxxxxxxxxxxxx'
SpotifyClientSecret='xxxxxxxxxxxxxxxxxxxxxxxx'

username = 'xxxxxxxxxxxxxxxxxxxxx'

scope = 'user-library-read,playlist-read-private,user-read-recently-played,playlist-modify-public,user-top-read'

token = util.prompt_for_user_token(username,
                                   scope,
                                   client_id=SpotifyClientID,
                                   client_secret=SpotifyClientSecret,
                                   redirect_uri='http://localhost/')

sp = spotipy.Spotify(auth=token)

sp.current_user()

## {'country': 'PL', 'display_name': 'Łukasz Prokulski', 'explicit_content': {'filter_enabled': False, 'filter_locked': False}, 'external_urls': {'spotify': 'https://open.spotify.com/user/1166639346'}, 'followers': {'href': None, 'total': 11}, 'href': 'https://api.spotify.com/v1/users/1166639346', 'id': '1166639346', 'images': [{'height': None, 'url': 'https://scontent.xx.fbcdn.net/v/t1.0-1/p320x320/72530210_10157567070828864_2506969271054106624_o.jpg?_nc_cat=101&_nc_sid=0c64ff&_nc_ohc=tOl_U1v3KMQAX_kD1P9&_nc_ht=scontent.xx&_nc_tp=6&oh=92cdcafca7a1ceaed0ef466723bf4711&oe=5E9FE65B', 'width': None}], 'product': 'premium', 'type': 'user', 'uri': 'spotify:user:1166639346'}

for i, track in enumerate(sp.current_user_top_tracks(limit=50, time_range='short_term')['items']):
    print(f"{i+1}. {track['id']} \"{track['name']}\" - {track['artists'][0]['name']} (LP \"{track['album']['name']}\")")


## 1. 0z1b34WikhOH9ZxU8QDWcv "One Headlight" - The Wallflowers (LP "Bringing Down The Horse")
## 2. 2EC9IJj7g0mN1Q5VrZkiYY "Rebel Rebel - 2016 Remaster" - David Bowie (LP "Diamond Dogs (2016 Remastered Version)")
## 3. 4pGqFOfzvfe6avb9kbZicC "Sweater Weather" - The Neighbourhood (LP "I'm Sorry...")
## 4. 60wxK3idlzc3lykP8cYZ2l "My Mistake" - Gabrielle Aplin (LP "Dear Happy")
## 5. 0uppYCG86ajpV2hSR3dJJ0 "Give It Away" - Red Hot Chili Peppers (LP "Blood Sugar Sex Magik (Deluxe Edition)")
## 6. 2aoo2jlRnM3A0NyLQqMN2f "All Along the Watchtower" - Jimi Hendrix (LP "Electric Ladyland")
## 7. 4ReyTz0y3TGkX48wO3Llot "Epic" - Faith No More (LP "The Real Thing")
## 8. 02acUMylPHMjJ9miDc9b38 "Ever the Same" - Rob Thomas (LP "Something To Be")
## 9. 05NYcsjJwOYq4jIiKPVj9p "Hard To Handle" - The Black Crowes (LP "Shake Your Money Maker")
## 10. 09nEPqqMCJ5gNioi7M2dth "New Soul" - Yael Naim (LP "Yael Naim")
## 11. 0leJYiYyZE6jjXCEVYUQAm "Heroin" - Badflower (LP "OK, I'M SICK")
## 12. 0tZ3mElWcr74OOhKEiNz1x "Bulls On Parade" - Rage Against The Machine (LP "Evil Empire")
## 13. 1TF8mSG5YHyaS4o3NeBSAo "Munich" - Editors (LP "The Back Room")
## 14. 2EaWqg4KUDVx6gVbgQ4Lyw "Przebój nocy" - PRO8L3M (LP "Art Brut 2")
## 15. 2fICdpdRotwfmGYzvs8Ngf "Rock You Like a Hurricane - 2015 - Remaster" - Scorpions (LP "Love At First Sting (Deluxe Edition)")
## 16. 4QbvmYYVLl8LNUsOXstLft "Take This Waltz - Paris Version" - Leonard Cohen (LP "I'm Your Man")
## 17. 4qO03RMQm88DdpTJcxlglY "Call Me" - Blondie (LP "Atomic/Atomix")
## 18. 52cfCOnS2x2TFuM0pFLl8x "Pigs on the Wing, Pt. 1 - Live" - Roger Waters (LP "In the Flesh - Live")
## 19. 5MMnwYs0hIxkENRsbkWJ2G "Smoke On The Water - Remastered 2012" - Deep Purple (LP "Machine Head (Remastered)")
## 20. 5OpDM7JORZzXPo0YIN3vYq "Tower of Song" - Leonard Cohen (LP "I'm Your Man")
## 21. 5Z8EDau8uNcP1E8JvmfkZe "School's Out" - Alice Cooper (LP "School's Out")
## 22. 5aT3i3rdNH74QWbA1o6qwl "Język Wszechświata" - Fisz (LP "Polepione Dźwięki")
## 23. 5anZi6BM5SWn6u6wrVYsgy "5 O'clock (K-Def Remix) [feat. Nonchalant] [Bonus Track]" - Pete Rock, Marley Marl (LP "Future Flavas Mixtape")
## 24. 5arVt2Wg0zbiWwAOZef2Nl "Higher Ground - Remastered" - Red Hot Chili Peppers (LP "Mother's Milk")
## 25. 5fztgDIt1Nq32VHJrAHq0Y "Big Yellow Taxi" - Counting Crows (LP "Hard Candy")
## 26. 6EPRKhUOdiFSQwGBRBbvsZ "Ace of Spades" - Motörhead (LP "Ace of Spades (Expanded Edition)")
## 27. 6MB1W09LUZCYrXTA0JdW3M "Red Wind" - Jan Garbarek (LP "Visible World")
## 28. 6QewNVIDKdSl8Y3ycuHIei "Even Flow" - Pearl Jam (LP "Ten")
## 29. 6dxfn2W3dN7MXWvnOJI3qS "Pygmy Lullaby" - Jan Garbarek (LP "Visible World")
## 30. 6hnkM0k52jzX3OLP9cgbuV "Ballad for Bernt" - Robert Majewski (LP "Plays Komeda")
## 31. 6y20BV5L33R8YXM0YuI38N "No One Knows" - Queens of the Stone Age (LP "Songs For The Deaf")
## 32. 7KPTwEENcZ5r4KJ6FHmhLP "Home" - Daughtry (LP "It's Not Over....The Hits So Far")
## 33. 14sAMwqbuFDSlLKmVBD9MI "Girls Chase Boys" - Ingrid Michaelson (LP "Lights Out")
## 34. 1yC4zXWJuywuVp5Hm5nobQ "You Shouldn't Do That - 1996 Remaster" - Hawkwind (LP "In Search Of Space")
## 35. 2aBxt229cbLDOvtL7Xbb9x "Always Be My Baby" - Mariah Carey (LP "Daydream")
## 36. 4m0q0xQ2BNl9SCAGKyfiGZ "Somebody Else" - The 1975 (LP "I like it when you sleep, for you are so beautiful yet so unaware of it")
## 37. 3pSJtP2BOA9zxU7r6bPPkH "In A Hand Or A Face" - The Who (LP "The Who By Numbers")
## 38. 2E5vl5eF4M4QbaADGegVDP "The Partisan" - Leonard Cohen (LP "Songs From A Room")
## 39. 2espysVKOqumSCzevPfdSa "Your Possible Pasts - 2011 Remastered Version" - Pink Floyd (LP "The Final Cut (2011 Remastered Version)")
## 40. 2hQADwe4Ve0ZHwBDrBEuTF "Pierwszy raz" - Dezerter (LP "Blasfemia")
## 41. 3ix6K4wZY29bCujrSznwFZ "Let's Dance - 2018 Remaster" - David Bowie (LP "Let's Dance (2018 Remaster)")
## 42. 3q4hJtKfWjpJrEFtSFmqAf "Master of the Universe - 1996 Remaster" - Hawkwind (LP "In Search Of Space")
## 43. 45Ia1U4KtIjAPPU7Wv1Sea "Are You Gonna Go My Way" - Lenny Kravitz (LP "Are You Gonna Go My Way")
## 44. 4NirsDE9nOl4HnwlZOIwR1 "Polepiony" - Fisz (LP "Polepione Dźwięki")
## 45. 56xbGPjb5lWowId6TzWYDP "One Of The Few - 2011 Remastered Version" - Pink Floyd (LP "The Final Cut (2011 Remastered Version)")
## 46. 5atFAXIgMXP378op4FaxBm "I'm Your Man" - Leonard Cohen (LP "I'm Your Man")
## 47. 5uF3hldjpeDjlSPhUYa1Ob "The Post War Dream - 2011 Remastered Version" - Pink Floyd (LP "The Final Cut (2011 Remastered Version)")
## 48. 6GtX0jaNL8IjVQfrDBx81z "Bullet With Butterfly Wings - Remastered 2012" - The Smashing Pumpkins (LP "Mellon Collie And The Infinite Sadness (Deluxe Edition)")
## 49. 6UEzGV76d96lYTA68iBO0B "First We Take Manhattan" - Leonard Cohen (LP "I'm Your Man")
## 50. 6d0LXn0CWYufMF3G6iJn14 "I Can't Forget" - Leonard Cohen (LP "I'm Your Man")

def get_track_info(track_id):

    # próbujemy pobrać informacje o konkretnym track id
    try:
        track_info = sp.track(track_id)
    except:
        return pd.DataFrame()

    # pobieramy cechy dla tego track id
    audio_features = sp.audio_features(track_id)

    # uwaga - czasem tych cech nie ma
    if audio_features[0] is not None:
        # jeśli wszystko udało się zgromadzić pakujemy w wiersz pandasowej tabeli
        tracks_df = pd.DataFrame({'track_id': track_info['id'],
                                 'track_title': track_info['name'],
                                 'track_artist': track_info['artists'][0]['name'],
                                 'album_id': track_info['album']['id'],
                                 'album_title': track_info['album']['name'],
                                 'album_artist': track_info['album']['artists'][0]['name'],
                                 'album_release_date': track_info['album']['release_date'],
                                 'track_popularity': track_info['popularity'],
                                 'explict': track_info['explicit'],
                                 'danceability': audio_features[0]['danceability'],
                                 'energy': audio_features[0]['energy'],
                                 'key': audio_features[0]['key'],
                                 'loudness': audio_features[0]['loudness'],
                                 'mode': audio_features[0]['mode'],
                                 'speechiness': audio_features[0]['speechiness'],
                                 'acousticness': audio_features[0]['acousticness'],
                                 'instrumentalness': audio_features[0]['instrumentalness'],
                                 'liveness': audio_features[0]['liveness'],
                                 'valence': audio_features[0]['valence'],
                                 'tempo': audio_features[0]['tempo'],
                                 'duration_ms': audio_features[0]['duration_ms'],
                                 'time_signature': audio_features[0]['time_signature']},
                                index=[0])

        # wydzielamy dodatkowo rok wydania z daty
        tracks_df['album_release_year'] = tracks_df['album_release_date'].apply(lambda d: pd.to_datetime(d).year)
    else:
         return pd.DataFrame()

    return(tracks_df)

tracks_df = pd.DataFrame()

tracks = sp.current_user_saved_tracks()

while tracks:
    for tr in tracks['items']:
        tracks_df = tracks_df.append(get_track_info(tr['track']['id']))

    tracks = sp.next(tracks)


print(tracks_df.head(10))

tracks_df.to_csv("spotify_loved_tracks.csv", index=False)

import matplotlib.pyplot as plt
import seaborn as sns

fig = plt.figure(figsize=(15,15))
sns.pairplot(tracks_df[['danceability', 'energy', 'key','loudness',
                        'mode', 'speechiness', 'acousticness',
                        'instrumentalness', 'liveness', 'valence',
                        'tempo', 'duration_ms', 'time_signature',
                        'album_release_year']],
             kind="reg")

plt.show()

spotify_playlist = sp.user_playlists('spotify')

for playlist in spotify_playlist['items']:
    print(f"\n{playlist['id']}: {playlist['name']} ({playlist['description']}) - {playlist['tracks']['total']} tracks:")
    for i, track in enumerate(sp.playlist(playlist['id'])['tracks']['items']):
        if track['track'] is not None:
            print(f"\t{i}\t{track['track']['id']} {track['track']['name']} - {track['track']['artists'][0]['name']}")

# lista ID playlist, z utworami które mi się nie podobają
bad_playlists = ['37i9dQZF1DXcBWIGoYBM5M', '37i9dQZF1DX0XUsuxWHRQd', '37i9dQZF1DX1lVhptIYRda',
                 '37i9dQZF1DX10zKzsJ2jva', '37i9dQZF1DX4JAvHpjipBk', '37i9dQZF1DX4dyzvuaRJ0n',
                 '37i9dQZF1DX4SBhb3fqCJd', '37i9dQZF1DX4sWSpwq3LiO', '37i9dQZF1DWXJfnUiYjUKT',
                 '37i9dQZF1DXcRXFNfZr7Tp', '37i9dQZF1DWTwnEm1IYyoj', '37i9dQZF1DX2A29LI7xHn1',
                 '37i9dQZF1DX2RxBh64BHjQ', '37i9dQZF1DWVA1Gq4XHa6U', '37i9dQZF1DWY4xHQp97fN6',
                 '37i9dQZF1DWX3387IZmjNa', '37i9dQZF1DWYkaDif7Ztbp', '37i9dQZF1DX5hR0J49CmXC',
                 '37i9dQZF1DXan38dNVDdl4', '37i9dQZF1DWSvKsRPPnv5o', '37i9dQZF1DWUVpAXiEPK8P',
                 '37i9dQZF1DX0Tkc6ltcBfU', '37i9dQZF1DX1YPTAhwehsC', '37i9dQZF1DXaKctwWdt4be',
                 '37i9dQZF1DWTggY0yqBxES', '37i9dQZF1DX0HRj9P7NxeE', '37i9dQZF1DWT6SJaitNDax',
                 '37i9dQZF1DWZkHEX2YHpDV']

# tutaj będziemy zbierać wszystkie informacje o utworach
bad_tracks_df = pd.DataFrame()

# dla każdej ze "złych" playlist:
for i, one_playlist in enumerate(bad_playlists):
    print(i, len(bad_playlists), one_playlist)

    # pobieramy listę utworów
    track_list = sp.playlist(one_playlist)

    # dla każdego utworu
    for trid in track_list['tracks']['items']:
        # pobieramy cechy
        if trid['track']:
            bad_tracks_df = bad_tracks_df.append(get_track_info(trid['track']['id']))

# usuwamy duplikaty
bad_tracks_df.drop_duplicates(inplace=False)

#usuwamy te utwory, których wykonwcy są w ulubionych
bad_tracks_df = bad_tracks_df[~bad_tracks_df['album_artist'].isin(tracks_df['album_artist'].unique())]

# zapisujemy do pliku
bad_tracks_df.to_csv("spotify_heted_tracks.csv", index=False)

print(bad_tracks_df.head(10))

fig = plt.figure(figsize=(15,15))
sns.pairplot(bad_tracks_df[['danceability', 'energy', 'key','loudness',
                        'mode', 'speechiness', 'acousticness',
                        'instrumentalness', 'liveness', 'valence',
                        'tempo', 'duration_ms', 'time_signature',
                        'album_release_year']],
             kind="reg")

plt.show()

from sklearn.model_selection import train_test_split

# oznaczamy ulubione piosenki klasą "1"
tracks_df['liked'] = 1

# a nielubiane klasą "0"
bad_tracks_df['liked'] = 0

# z nielubianych bierzemy tyle ile jest lubianych (bo akurat wiemy że nielubianych jest więcej)
disliked_df = bad_tracks_df.sample(n=tracks_df.shape[0])

# cały zbiór
all_tracks_full = tracks_df.append(disliked_df).drop_duplicates()

# przemieszajmy go dla pewności
all_tracks_full = all_tracks_full.sample(frac=1).reset_index(drop=True)

# i tylko numeryczne kolumny
all_tracks = all_tracks_full[['track_id', 'track_popularity',
       'danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness',
       'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
       'duration_ms', 'time_signature', 'album_release_year', 'liked']]


# podział na trening i test
X_train, X_test, y_train, y_test = train_test_split(all_tracks.drop('liked', axis=1),
                                                    all_tracks['liked'],
                                                    test_size=0.2)

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, confusion_matrix

# budujemy model i uczymy go
model = RandomForestClassifier()
model.fit(X_train.drop('track_id', axis=1), y_train)

# obliczamy predykcje dla danych testowych
y_pred = model.predict(X_test.drop('track_id', axis=1))

# jaki mamy wynik?
print(classification_report(y_test, y_pred))


##               precision    recall  f1-score   support
##
##            0       0.96      0.98      0.97       223
##            1       0.98      0.97      0.97       276
##
##     accuracy                           0.97       499
##    macro avg       0.97      0.97      0.97       499
## weighted avg       0.97      0.97      0.97       499

X_test['predicted_like'] = y_pred

# z przewidzianych wyników zostawiamy tylko to, co ma się potencjalnie podobać
new_tracks = X_test[X_test['predicted_like'] == 1][['track_id']]

# dodajemy info o wykonawcy i tym czy już było lajkowane
new_tracks = new_tracks.merge(all_tracks_full, how='inner', on='track_id')

# i usuwamy to co już miało lajka
new_tracks = new_tracks[new_tracks['liked'] == 0]

# zostawiamy tylko potrzebne kolumny
new_tracks = new_tracks[['track_id', 'track_title', 'track_artist']]

# wynik:
print(new_tracks)

import time

# tworzymy nową playlistę - nazwa z czasem utworzenia
new_playlist = sp.user_playlist_create(username, f"Lista z #ML {time.strftime('%Y-%m-%d %H:%M')}", public=True)

# dodajemy do niej piosneczki
res = sp.user_playlist_add_tracks(username, new_playlist['id'], new_tracks.track_id.values)

for i, a in enumerate(sp.search("The Wall Pink Floyd", type='album')['albums']['items']):
    print(i+1, a['id'], a['artists'][0]['name'], a['name'], a['release_date'], a['total_tracks'])


## 1 6WaIQHxEHtZL0RZ62AuY0g Pink Floyd The Wall (Remastered) 1979-11-30 26
## 2 4Y8K5Cfs0vdi5M2RZSF1Pp Vitamin String Quartet VSQ Performs Pink Floyd's The Wall: More Bricks 2006-02-21 12
## 3 2ZQXrdTaiUvgbr22cvgy6D Gustavo Zavala The Wall for Babies: The Complete Pink Floyd-The Wall Album for Babies 2012-12-03 26
## 4 5ISax6a3D5kBQKY1zeFsYA Atom Pink Floyd Tribute Tearing Down the Wall: Live 2015 2016-12-05 10
## 5 6273GngZvwpRWgg99GsESp Various Artists Re-Building The Wall - A Tribute To Pink Floyd 2007-01-01 26
## 6 42s70vpGM4iOk0BWhhpsVe Various Artists Back Against the Wall - A Tribute to Pink Floyd (Bonus Track Version) 2015-01-20 31
## 7 1aHDf1VsnUKbMPJX6RtwLH Various Artists Back Against The Wall – A Tribute To Pink Floyd 2005-11-15 26
## 8 6mntef7MKxZKqEL2Aw3RXb Done Again Another Brick in the Wall (Part II) [In the Style of Pink Floyd] {Performance Track with Demonstration Vocals} - Single 2012-06-13 1
## 9 12WlibQIazyVIlduAqBB15 Solo Sounds Solo Cello: Pink Floyd's the Wall 2017-04-14 27
## 10 6ijyx2AhwsHASUIJjdOu6L Rockopera Tribute to Pink Floyd: The Wall Live Orchestra 2013-04-20 25

for i, t in enumerate(sp.album_tracks('6WaIQHxEHtZL0RZ62AuY0g')['items']):
    print(i+1, t['id'], t['name'])


## 1 1KaIRTf1YEWt54V3IW34qV In The Flesh? - 2011 Remastered Version
## 2 3dHzjlxUULSWDsTHFc2zM7 The Thin Ice - 2011 Remastered Version
## 3 4j2bQkwIEO7QzAdjJ4txVm Another Brick In The Wall, Pt. 1 - 2011 Remastered Version
## 4 6ASehhdN08JbXzDJmMrogn The Happiest Days Of Our Lives - 2011 Remastered Version
## 5 7rPzEczIS574IgPaiPieS3 Another Brick In The Wall, Pt. 2 - 2011 Remastered Version
## 6 5xdySZ8v0lEOKblcwXWT3M Mother - 2011 Remastered Version
## 7 4Y1Ra2qfmTYOugWzBfocAX Goodbye Blue Sky - 2011 Remastered Version
## 8 6tsfGbVHWirffWWUUmtalg Empty Spaces - 2011 Remastered Version
## 9 4ZhPXTpqp3NojMWsNHGnSd Young Lust - 2011 Remastered Version
## 10 0MW7g3oFkdmZIiomSGC4x4 One Of My Turns - 2011 Remastered Version
## 11 3wm0FEw4BhPvCnSK8L0vGP Don't Leave Me Now - 2011 Remastered Version
## 12 22IiK3b7NSBfmUrnhj7rSr Another Brick In The Wall, Pt. 3 - 2011 Remastered Version
## 13 3mPeMKup50f1llqTciEfLJ Goodbye Cruel World - 2011 Remastered Version
## 14 5UJmwdqGP7RONuVzYnHjUp Hey You - 2011 Remastered Version
## 15 2rKcC5vlTsDM94RWTPoyHV Is There Anybody Out There? - 2011 Remastered Version
## 16 3kM50GD76NgPWaTfKKKTz3 Nobody Home - 2011 Remastered Version
## 17 2rzHK6y5GTxaz8104rp2cA Vera - 2011 Remastered Version
## 18 4rvDQAW3boZffZIzaYqnAc Bring The Boys Back Home - 2011 Remastered Version
## 19 7Fg4jpwpkdkGCvq1rrXnvx Comfortably Numb - 2011 Remastered Version
## 20 7myPwy9eGrRv9jNAoeTh2s The Show Must Go On - 2011 Remastered Version
## 21 0B545eDyCppapUQ1ujrTmg In The Flesh - 2011 Remaster
## 22 7CLHw352I29cIRf944RQnO Run Like Hell - 2011 Remastered Version
## 23 7s9MWRmBH5fbohw3Tqikyb Waiting For The Worms - 2011 Remastered Version
## 24 6X9k4QxABkn8luC8P93OZf Stop - 2011 Remastered Version
## 25 7p8QuQC5aIRY3q9wWE2f6T The Trial - 2011 Remastered Version
## 26 1zeyA74WeKOpYH6wtXQwkp Outside The Wall - 2011 Remastered Version

# bierzemy listę ID utworów
tracks_to_score_ids = [trid['id'] for trid in sp.album_tracks('6WaIQHxEHtZL0RZ62AuY0g')['items']]

tracks_to_score = pd.DataFrame()

# dla każdego utworu podobieramy o nim info
for track in tracks_to_score_ids:
    tracks_to_score = tracks_to_score.append(get_track_info(track))

tracks_to_score['liked'] = [True if tid in list(tracks_df['track_id']) else False for tid in tracks_to_score['track_id'] ]

print(tracks_to_score[['track_title', 'track_popularity', 'liked']])

# nie wszystkie kolumny są używane w modelu
X_tracks_to_score = tracks_to_score[['track_popularity',
       'danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness',
       'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
       'duration_ms', 'time_signature', 'album_release_year']]

# uwaga - potrzebujemy tutaj prawdopodobieństwo przynależności do klasy (i bierzemy tylko to, które dotyczy "jedynki") a nie samą klasę
tracks_to_score['like_prob'] = model.predict_proba(X_tracks_to_score)[:, 1]

# jaki mamy wynik? z jakim prawdopodobieństwem piosenka mi się spodoba?
print(tracks_to_score[['track_title', 'liked', 'like_prob']])

def score_album_trakcs(album_id):
    tracks_to_score_ids = [trid['id'] for trid in sp.album_tracks(album_id)['items']]

    tracks_to_score = pd.DataFrame()

    for track in tracks_to_score_ids:
        tracks_to_score = tracks_to_score.append(get_track_info(track))

    # to w sumie niczego nie wnosi, ale będzie widać, że nie było żadnego lajka na tej płycie
    tracks_to_score['liked'] = [True if tid in list(tracks_df['track_id']) else False for tid in tracks_to_score['track_id'] ]

    X_tracks_to_score = tracks_to_score[['track_popularity',
           'danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness',
           'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
           'duration_ms', 'time_signature', 'album_release_year']]

    tracks_to_score['like_prob'] = model.predict_proba(X_tracks_to_score)[:, 1]

    return tracks_to_score[['track_title', 'liked', 'like_prob']]


print(score_album_trakcs('3p7WXDBxhC5KS9IFXnwae7'))

print(score_album_trakcs('5C0YLr4OoRGFDaqdMQmkeH'))

print(score_album_trakcs('6R8ir7mqaqV3vrtv5pmvID'))

print(score_album_trakcs('7xV2TzoaVc0ycW7fwBwAml'))