Sparsh Gupta and Sohum Kothavade
In this project, we explore how certain non-musical factors, for example, release date of songs, album, etc. are correlated to the 'popularity' of a song and how these factors affect the successfulness of a song. To do this, we used Spotify's API to obtain data from 'The Million Playlist' dataset. We process this data to extract statistics for the non-musical factors being considered in this study and use visualizations to understand how these are related to the 'popularity' of a song.
For understanding this project thouroughly, please access the computational essay
spotify-hitsongs.ipynb included in this repository.
| Package | Uses |
|---|---|
| Matplotlib | Visualization |
| NumPy | Random Sampling |
| Pandas | Data Processing |
| Requests | Web Requests |
| SpotiPy | Spotify API |
The dependencies are present in requirements.txt and can be installed using the following in terminal/command prompt (make sure to have your present working directory as this repo):
pip install -r requirements.txt
The data used in this study is from Spotify's 'The Million Playlist' dataset. We obtained the data used in this study from Spotify's API which gives us access to a million playlists of songs consisting of more than a hundred million songs/tracks.
The sample extracted data output can be seen below:
We obtain the web API access keys which includes CLIENT_ID and CLIENT_SECRET from Spotify Developer Tools by creating a developer account and a project.
If you would like to authorize into your Spotify User Account before extracting the API data (recommended), you can use the module spotifyrequests.py and enter your CLIENT_ID and CLIENT_SECRET. If the request to authorize the account succeeds, you will be outputted an alphanumeric API access token. If you want to access the API data without user authorization (not recommended), you can skip this step.
Once you obtain the CLIENT_ID and CLIENT_SECRET, enter the keys in spotify_api_keys.py inside the appropriate functions
We use the following functions from spotify_api_keys.py (private file) to access the API keys stored in this via the following functions:
CLIENT_ID = get_client_id()
CLIENT_SECRET = get_client_secret()
To run the code in this repo, please clone this repo to your local machine and run main.py in either a Python-compatible IDE or if using a terminal/command prompt (make sure to have your present working directory as this repo):
python3 main.py
Executing main.py using python will automatically extract the dataset, process it, and generate visualizations, provided the above API Access step is completed successfully.
We used the module spotifydata.py to extract data from Spotify's web API using the following function.
get_spotify_data(CLIENT_ID, CLIENT_SECRET)
The above function also processes the dataset accessed through the API and stores it to the csv file data.csv
The visualizations for this project are created using the following function from the module visualizations.py
create_visualizations('data.csv', 'figures')
The function uses the data file data.csv to generate visualizations and stores the visualizations in the directory figures
The unit tests for this project tests the code for functions get_spotify_data() and create_visualizations() in the module spotify_hit_songs_test.py