05: Project¶

While brainstorming about what project could we work upon that would allow us to learn and revise through the concepts we have learned throughout the bootcamp, and at the same time make it really cool, we came up with the idea of creating a Machine Learning Model based on data from Spotify!

Through this session (the last one😔), we'll be creating k-NN based ML model which allow us to classify songs into particular playlists.

Okay enough talk, let's get started! 🎵

1: Accessing Spotify Credentials¶

Before we get started, we'll need a source for our data for this project. As mentioned before, we'll be fetching the data we need, from Spotify itself!

To do so, first create a Spotify account.

Once done creating an account, head on to "Spotify for Developers", and navigate to "Dashboard".

After opening up the Dashboard, click on Create an App, with which you'll get Client ID and Client Secret (which will be used in the code we'll be writing).

Get Spotify Client Information¶

In [1]:

!pip install spotipy

!pip install spotipy

Requirement already satisfied: spotipy in /usr/local/lib/python3.7/dist-packages (2.19.0)
Requirement already satisfied: urllib3>=1.26.0 in /usr/local/lib/python3.7/dist-packages (from spotipy) (1.26.9)
Requirement already satisfied: six>=1.15.0 in /usr/local/lib/python3.7/dist-packages (from spotipy) (1.15.0)
Requirement already satisfied: requests>=2.25.0 in /usr/local/lib/python3.7/dist-packages (from spotipy) (2.27.1)
Requirement already satisfied: charset-normalizer~=2.0.0 in /usr/local/lib/python3.7/dist-packages (from requests>=2.25.0->spotipy) (2.0.12)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests>=2.25.0->spotipy) (2021.10.8)
Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests>=2.25.0->spotipy) (2.10)

In [2]:

import spotipy
import pandas as pd
from spotipy.oauth2 import SpotifyClientCredentials
# client_id = "<ENTER CLIENT ID>"
client_id = '93afa71bdab647d3a971688255f7fcaf'
# client_secret = "<ENTER CLIENT SECRET"
client_secret = '0f834b4581174c849c5a4eb962c74ac8'
client_credentials_manager = SpotifyClientCredentials('93afa71bdab647d3a971688255f7fcaf', '0f834b4581174c849c5a4eb962c74ac8')
# client_credentials_manager = SpotifyClientCredentials('<ENTER SPOTIFY USERNAME>', '<ENTER CLIENT SECRET>')
sp = spotipy.Spotify(client_credentials_manager = client_credentials_manager)

import spotipy import pandas as pd from spotipy.oauth2 import SpotifyClientCredentials # client_id = "" client_id = '93afa71bdab647d3a971688255f7fcaf' # client_secret = "', '') sp = spotipy.Spotify(client_credentials_manager = client_credentials_manager)

Make sure to enter Client ID, Client Secret, and Username at the positions mentioned above in the code.

Download playlists and convert into CSV¶

In [3]:

id_test = sp.user_playlist_tracks('317d3qqu6wx3z2l444u2vde6onde', 'spotify:playlist:37i9dQZF1DXdPec7aLTmlC')['items'][0]['track']['id']
# id_test = sp.user_playlist_tracks('<ENTER SPOTIFY USERNAME>', '<ENTER SPOTIFY URI>')['items'][0]['track']['id']
columns = ['artist', 'track']
list(map(lambda x: columns.append(x), list(sp.audio_features(id_test)[0].keys())))

playlist_tracks = pd.DataFrame(columns = columns, index = range(0, 200))
playlist_tracks.head()

id_test = sp.user_playlist_tracks('317d3qqu6wx3z2l444u2vde6onde', 'spotify:playlist:37i9dQZF1DXdPec7aLTmlC')['items'][0]['track']['id'] # id_test = sp.user_playlist_tracks('', '')['items'][0]['track']['id'] columns = ['artist', 'track'] list(map(lambda x: columns.append(x), list(sp.audio_features(id_test)[0].keys()))) playlist_tracks = pd.DataFrame(columns = columns, index = range(0, 200)) playlist_tracks.head()

Out[3]:

	artist	track	danceability	energy	key	loudness	mode	speechiness	acousticness	instrumentalness	liveness	valence	tempo	type	id	uri	track_href	analysis_url	duration_ms	time_signature
0	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
1	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
2	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
3	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN
4	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

In [4]:

# Insert Spotify URIs into the list below to create a dataframe from the playlists
playlist_ids = ['spotify:playlist:1llkez7kiZtBeOw5UjFlJq', 'spotify:playlist:1h90L3LP8kAJ7KGjCV2Xfd']
row_counter = 0

for playlist_id in playlist_ids:
    for track in sp.user_playlist_tracks('317d3qqu6wx3z2l444u2vde6onde', playlist_id)['items']:
        current_id = track['track']['id']
        current_row = [track['track']['artists'][0]['name'], track['track']['name']]
        (list(map(lambda x: current_row.append(x), list(sp.audio_features(current_id)[0].values()))))
        playlist_tracks.iloc[row_counter] = current_row
        row_counter += 1

# Insert Spotify URIs into the list below to create a dataframe from the playlists playlist_ids = ['spotify:playlist:1llkez7kiZtBeOw5UjFlJq', 'spotify:playlist:1h90L3LP8kAJ7KGjCV2Xfd'] row_counter = 0 for playlist_id in playlist_ids: for track in sp.user_playlist_tracks('317d3qqu6wx3z2l444u2vde6onde', playlist_id)['items']: current_id = track['track']['id'] current_row = [track['track']['artists'][0]['name'], track['track']['name']] (list(map(lambda x: current_row.append(x), list(sp.audio_features(current_id)[0].values())))) playlist_tracks.iloc[row_counter] = current_row row_counter += 1

In [5]:

playlist_tracks.head()

playlist_tracks.head()

Out[5]:

	artist	track	danceability	energy	key	loudness	mode	speechiness	acousticness	instrumentalness	liveness	valence	tempo	type	id	uri	track_href	analysis_url	duration_ms	time_signature
0	Katrina & The Waves	Walking On Sunshine	0.596	0.869	10	-11.97	1	0.037	0.0116	0.173	0.0678	0.944	109.902	audio_features	05wIrZSwuaVWhcv5FfqeH0	spotify:track:05wIrZSwuaVWhcv5FfqeH0	https://api.spotify.com/v1/tracks/05wIrZSwuaVW...	https://api.spotify.com/v1/audio-analysis/05wI...	238733	4
1	WALK THE MOON	Shut Up and Dance	0.578	0.866	1	-3.804	1	0.0619	0.00701	0	0.257	0.619	128.038	audio_features	4kbj5MwxO1bq9wjT5g9HaA	spotify:track:4kbj5MwxO1bq9wjT5g9HaA	https://api.spotify.com/v1/tracks/4kbj5MwxO1bq...	https://api.spotify.com/v1/audio-analysis/4kbj...	199080	4
2	The Edwin Hawkins Singers	Oh Happy Day	0.321	0.462	8	-10.711	1	0.0346	0.0961	0	0.102	0.457	116.782	audio_features	34yIb7hw5SuR80au1BHTtt	spotify:track:34yIb7hw5SuR80au1BHTtt	https://api.spotify.com/v1/tracks/34yIb7hw5SuR...	https://api.spotify.com/v1/audio-analysis/34yI...	308600	4
3	Andy Grammer	Good To Be Alive (Hallelujah)	0.784	0.882	5	-3.646	1	0.0451	0.0377	0	0.65	0.699	120.017	audio_features	1fMEUGs5CvRv0VrcJlQITt	spotify:track:1fMEUGs5CvRv0VrcJlQITt	https://api.spotify.com/v1/tracks/1fMEUGs5CvRv...	https://api.spotify.com/v1/audio-analysis/1fME...	189870	4
4	American Authors	Best Day Of My Life	0.672	0.901	2	-2.415	1	0.0353	0.0545	0.000285	0.0792	0.507	99.989	audio_features	5k31cAfP8yymw8v0AYQIjr	spotify:track:5k31cAfP8yymw8v0AYQIjr	https://api.spotify.com/v1/tracks/5k31cAfP8yym...	https://api.spotify.com/v1/audio-analysis/5k31...	194000	4

In [6]:

playlist_tracks.to_csv('happy.csv', encoding='utf-8', index = False)

playlist_tracks.to_csv('happy.csv', encoding='utf-8', index = False)

In [7]:

# playlist_tracks.to_csv(r'D:\Causality\GitHub\spotify-jam\happy_all.csv', index = False, header = True)

# playlist_tracks.to_csv(r'D:\Causality\GitHub\spotify-jam\happy_all.csv', index = False, header = True)

Similarly, create a csv file of calm songs.

Import the required libraries¶

In [8]:

from sklearn.neighbors import NearestNeighbors
import pandas as pd
import plotly.express as px
import matplotlib.pyplot as plt
get_ipython().run_line_magic('matplotlib', 'inline')
import numpy as np
import seaborn as sns

from sklearn.neighbors import NearestNeighbors import pandas as pd import plotly.express as px import matplotlib.pyplot as plt get_ipython().run_line_magic('matplotlib', 'inline') import numpy as np import seaborn as sns

Read the CSV files¶

In [9]:

happy = pd.read_csv("happy.csv")
calm = pd.read_csv("calm.csv")

happy = pd.read_csv("happy.csv") calm = pd.read_csv("calm.csv")

In [10]:

happy.head()

happy.head()

Out[10]:

	artist	track	danceability	energy	key	loudness	mode	speechiness	acousticness	instrumentalness	liveness	valence	tempo	type	id	uri	track_href	analysis_url	duration_ms	time_signature
0	Katrina & The Waves	Walking On Sunshine	0.596	0.869	10.0	-11.970	1.0	0.0370	0.01160	0.173000	0.0678	0.944	109.902	audio_features	05wIrZSwuaVWhcv5FfqeH0	spotify:track:05wIrZSwuaVWhcv5FfqeH0	https://api.spotify.com/v1/tracks/05wIrZSwuaVW...	https://api.spotify.com/v1/audio-analysis/05wI...	238733.0	4.0
1	WALK THE MOON	Shut Up and Dance	0.578	0.866	1.0	-3.804	1.0	0.0619	0.00701	0.000000	0.2570	0.619	128.038	audio_features	4kbj5MwxO1bq9wjT5g9HaA	spotify:track:4kbj5MwxO1bq9wjT5g9HaA	https://api.spotify.com/v1/tracks/4kbj5MwxO1bq...	https://api.spotify.com/v1/audio-analysis/4kbj...	199080.0	4.0
2	The Edwin Hawkins Singers	Oh Happy Day	0.321	0.462	8.0	-10.711	1.0	0.0346	0.09610	0.000000	0.1020	0.457	116.782	audio_features	34yIb7hw5SuR80au1BHTtt	spotify:track:34yIb7hw5SuR80au1BHTtt	https://api.spotify.com/v1/tracks/34yIb7hw5SuR...	https://api.spotify.com/v1/audio-analysis/34yI...	308600.0	4.0
3	Andy Grammer	Good To Be Alive (Hallelujah)	0.784	0.882	5.0	-3.646	1.0	0.0451	0.03770	0.000000	0.6500	0.699	120.017	audio_features	1fMEUGs5CvRv0VrcJlQITt	spotify:track:1fMEUGs5CvRv0VrcJlQITt	https://api.spotify.com/v1/tracks/1fMEUGs5CvRv...	https://api.spotify.com/v1/audio-analysis/1fME...	189870.0	4.0
4	American Authors	Best Day Of My Life	0.672	0.901	2.0	-2.415	1.0	0.0353	0.05450	0.000285	0.0792	0.507	99.989	audio_features	5k31cAfP8yymw8v0AYQIjr	spotify:track:5k31cAfP8yymw8v0AYQIjr	https://api.spotify.com/v1/tracks/5k31cAfP8yym...	https://api.spotify.com/v1/audio-analysis/5k31...	194000.0	4.0

Graph the attributes¶

In [11]:

plt.plot(happy.acousticness, color='blue')
plt.plot(calm.acousticness, color='red')
plt.xlabel("song #")
plt.ylabel("acousticness")
plt.title("acousticness in the playlists")

plt.plot(happy.acousticness, color='blue') plt.plot(calm.acousticness, color='red') plt.xlabel("song #") plt.ylabel("acousticness") plt.title("acousticness in the playlists")

Out[11]:

Text(0.5, 1.0, 'acousticness in the playlists')

Combining the datasets¶

In [12]:

frames = [happy, calm]

# Add target variable
for i in range(len(frames)):
  frames[i]['target'] = [i for _ in range(happy.shape[0])]

happy_calm = pd.concat(frames, ignore_index=True)
happy_calm = happy_calm[['energy', 'acousticness', 'valence', 'target']]

for col in ['energy', 'acousticness', 'valence']:
  happy_calm[col] = 1000 * happy_calm[col]

happy_calm.to_csv("happy_calm.csv", encoding='utf-8', index = False)

frames = [happy, calm] # Add target variable for i in range(len(frames)): frames[i]['target'] = [i for _ in range(happy.shape[0])] happy_calm = pd.concat(frames, ignore_index=True) happy_calm = happy_calm[['energy', 'acousticness', 'valence', 'target']] for col in ['energy', 'acousticness', 'valence']: happy_calm[col] = 1000 * happy_calm[col] happy_calm.to_csv("happy_calm.csv", encoding='utf-8', index = False)

In [13]:

happy_calm = pd.read_csv("happy_calm.csv")
happy_calm.head()

happy_calm = pd.read_csv("happy_calm.csv") happy_calm.head()

Out[13]:

	energy	acousticness	valence	target
0	869.0	11.60	944.0	0
1	866.0	7.01	619.0	0
2	462.0	96.10	457.0	0
3	882.0	37.70	699.0	0
4	901.0	54.50	507.0	0

Check for possible errors¶

In [14]:

# Null values
happy_calm.isnull().values.any()

# Null values happy_calm.isnull().values.any()

Out[14]:

True

In [15]:

# Check the datatypes
happy_calm.dtypes

# Check the datatypes happy_calm.dtypes

Out[15]:

energy          float64
acousticness    float64
valence         float64
target            int64
dtype: object

In [16]:

# Change the datatype
happy_calm = happy_calm.fillna(0)
for col in list(happy_calm.columns):
  happy_calm[col] = happy_calm[col].astype(int)
happy_calm

# Change the datatype happy_calm = happy_calm.fillna(0) for col in list(happy_calm.columns): happy_calm[col] = happy_calm[col].astype(int) happy_calm

Out[16]:

	energy	acousticness	valence	target
0	869	11	944	0
1	866	7	619	0
2	462	96	457	0
3	882	37	699	0
4	901	54	507	0
...	...	...	...	...
395	577	4	98	1
396	321	61	181	1
397	942	1	951	1
398	461	138	114	1
399	648	10	326	1

400 rows × 4 columns

Standardize the data¶

In [17]:

from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
scaler.fit(happy_calm.drop('target', axis=1))

from sklearn.preprocessing import StandardScaler scaler = StandardScaler() scaler.fit(happy_calm.drop('target', axis=1))

Out[17]:

StandardScaler()

In [18]:

scaled_features = scaler.transform(happy_calm.drop('target',axis=1))
scaled_features

scaled_features = scaler.transform(happy_calm.drop('target',axis=1)) scaled_features

Out[18]:

array([[ 1.23411859, -0.73616408,  1.71891449],
       [ 1.22337857, -0.75258692,  0.64184329],
       [-0.22294506, -0.38717872,  0.10496473],
       ...,
       [ 1.49545925, -0.77722118,  1.74211294],
       [-0.22652507, -0.2147389 , -1.03175964],
       [ 0.44293661, -0.74026979, -0.32917781]])

In [19]:

happy_calm_feat = pd.DataFrame(scaled_features, columns = happy_calm.columns[:-1])
happy_calm_feat.head()

happy_calm_feat = pd.DataFrame(scaled_features, columns = happy_calm.columns[:-1]) happy_calm_feat.head()

Out[19]:

	energy	acousticness	valence
0	1.234119	-0.736164	1.718914
1	1.223379	-0.752587	0.641843
2	-0.222945	-0.387179	0.104965
3	1.280659	-0.629416	0.906969
4	1.348679	-0.559619	0.270668

Split the data¶

In [20]:

from sklearn.model_selection import train_test_split
X = happy_calm_feat
y = happy_calm['target']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=30, shuffle=True)

from sklearn.model_selection import train_test_split X = happy_calm_feat y = happy_calm['target'] X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=30, shuffle=True)

Train the model¶

In [21]:

# Finding k value
import math
math.sqrt(len(y_test))

# Finding k value import math math.sqrt(len(y_test))

Out[21]:

10.954451150103322

In [22]:

from sklearn.neighbors import KNeighborsClassifier

knn = KNeighborsClassifier(n_neighbors=7,p=2,metric='euclidean')
knn.fit(X_train, y_train)

from sklearn.neighbors import KNeighborsClassifier knn = KNeighborsClassifier(n_neighbors=7,p=2,metric='euclidean') knn.fit(X_train, y_train)

Out[22]:

KNeighborsClassifier(metric='euclidean', n_neighbors=7)

Make predictions¶

In [23]:

prediction = knn.predict(X_test)
prediction

prediction = knn.predict(X_test) prediction

Out[23]:

array([1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0,
       0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1, 0, 1, 0, 0, 0, 1, 1, 1,
       0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0,
       1, 1, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0,
       0, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1,
       1, 0, 1, 1, 0, 0, 1, 1, 0, 0])

Evaluate Predictions¶

In [24]:

from sklearn.metrics import classification_report, confusion_matrix
print(classification_report(y_test, prediction))

from sklearn.metrics import classification_report, confusion_matrix print(classification_report(y_test, prediction))

              precision    recall  f1-score   support

           0       0.85      0.83      0.84        63
           1       0.81      0.84      0.83        57

    accuracy                           0.83       120
   macro avg       0.83      0.83      0.83       120
weighted avg       0.83      0.83      0.83       120

Test other k values¶

In [25]:

error_rate = []
for i in range(1,40):
    
    knn = KNeighborsClassifier(n_neighbors=i)
    knn.fit(X_train, y_train)
    prediction_i = knn.predict(X_test)
    error_rate.append(np.mean(prediction_i != y_test))

error_rate = [] for i in range(1,40): knn = KNeighborsClassifier(n_neighbors=i) knn.fit(X_train, y_train) prediction_i = knn.predict(X_test) error_rate.append(np.mean(prediction_i != y_test))

Plot Error Rate¶

In [26]:

plt.plot(error_rate, color='blue')
plt.title('error rate vs. k values')
plt.xlabel('k values')
plt.ylabel('error rate')

plt.plot(error_rate, color='blue') plt.title('error rate vs. k values') plt.xlabel('k values') plt.ylabel('error rate')

Out[26]:

Text(0, 0.5, 'error rate')

Use the best k value¶

In [27]:

knn = KNeighborsClassifier(n_neighbors=7)
knn.fit(X_train, y_train)
prediction= knn.predict(X_test)
prediction

knn = KNeighborsClassifier(n_neighbors=7) knn.fit(X_train, y_train) prediction= knn.predict(X_test) prediction

Out[27]:

array([1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0,
       0, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 1, 0, 1, 0, 1, 0, 0, 0, 1, 1, 1,
       0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0,
       1, 1, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0, 1, 1, 1, 0, 1, 0, 1, 0, 1, 0,
       0, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1,
       1, 0, 1, 1, 0, 0, 1, 1, 0, 0])

In [28]:

print(classification_report(y_test, prediction))

print(classification_report(y_test, prediction))

              precision    recall  f1-score   support

           0       0.85      0.83      0.84        63
           1       0.81      0.84      0.83        57

    accuracy                           0.83       120
   macro avg       0.83      0.83      0.83       120
weighted avg       0.83      0.83      0.83       120

What more can I do?¶

There is so much you can do now with just this project!

Here are few suggestions that you folks must definitely try out:

• Adding more genres/emotions
• Automate creating playlists from the results
• Test it out with bigger datasets
• Improve the accuracy (try other Models that you have learned)
• Deploy it in the cloud
• Host the project in your GitHub profile
• And so much more! 🚀

The End?¶

Congratulations for completing the Machine Learning Bootcamp with ACM! 🎉

Pat yourself! You deserve it! You have done great throughout the bootcamp.

But is this the end? Haha! Don't worry, your journey through Machine Learning has only just started. You have an entire world that you can explore and create amazing projects with (one that could even change the way that we live our lives). All the best for this journey! 🙌

We, at ACM, know you'll ace it, like always! 💪