Spotify taste visualization

Introduction

A simple dashboard to observe the change in music taste. This project leverages the API provided by spotify.
I am not allowed to copy/distribute any data nor able to share my developer credentials.

Whoever whants to use or enhance this code can apply for credentials here: https://developer.spotify.com

Running

Changes in main

Change credentials in the dataImport script to your own:

credentials = {"client_id": **?**,
               "client_secret":**?*}

Change the username to the user of interest

username = **?**

Execution of the programm

Open the command window and change the directory to the location of the project. Use the following command to execute the dashboard:

bokeh serve spotify-taste-visualization

Navigate to the URL in your browser:

http://localhost:5006/spotify-taste-visualization

A new spotify login page will open and ask for your permission. Login and and allow your application the access to your data.

You will get redirected to a new URL. Copy and paste the url into your command window. The application is now running. Refresh the local http://localhost:5006/spotify-taste-visualization and start using the dashboard.

Explanation

Basic functionality

The dashboard visualizes the music library of the user. It consists of two main graphs. One is a scatter-plot of multiple song features with changing point sizes depending on a third feature. The second offers an aggregated overview of the new songs over time.

The user can decide between two data collection methods:

1. All saved songs by a user getData(credentials, username, saved = True) (default)
2. All songs that are at least in one playlist getData(credentials, username, saved = False)

Some potential selections are:
x/y:

• danceability
• loudness
• accusticness
• instrumentalness
• energy

size:

• None
• loudness
• duration in ms

Advanced functionality

Both graphs are connected to the same data. Therefore, selection datapoints in one graph highlights and selects the connected datapoints in the other graph. This allowes for a more specific inspection, depending on the periode of interest.

Additionally, certain patterns might be confusing or some features are not intuitive. Therefore, this dashboard allowes to hover over a specific data-point to optain additional information to what song this point corresponds.

TODO

Potential future enhancements:

• Adding filter possibilities
• Adding new aggregation selection methods
• Adding aggregated selection summary
• Adding data table of selection
• Adding differing aggregation time frames