A very prominent feature of the dance music community is the frequent collaborations between artists, whether through original songs or through remixing other artists’ existing songs. Some artist’s impact on the genre can be positively correlated with the number of artists they collaborate with and this impact was something I wanted to quantify and show through a network graph.

To pull data for this project, I wrote a python script to scrape Spotify of all of the songs from a list of artists I define. The program then creates combinations of each artist on the songs (discarding duplicates) and saves them to a dataset. Lastly the program discards all artists with under 7 collaborations to reduce noise and also removes all artists whose genre does not fall into the house/dance genre. The resulting dataset is a list of ‘relations’ (between ‘Artist 1’ and ‘Artist 2’) is saved to a CSV file that can be uploaded to Gephi to create visualizations. The artists I chose to filter on are 17 artists that I frequently listen to that cover a wide range of micro-genres within the Dance genre. This introduces a bias to the dataset because it is not representative of the entire industry but rather a subset of artists I listen to and the communities they belong to.

First Draft

My first idea was to create a double layer radial graph to highlight the top 5 artists with the most collaborations. The goal was to show these artists’ reach across the industry, but without organizing by community a lot of the meaning was getting lost in the jumble. Additionally, since this original model is across all years there is no way to differentiate the evolution of certain communities or growth of certain artists. Although the visual itself is very captivating to me, I decided it would be best to find a different layout that would be able to better showcase the community connections and be a bit easier to digest as a user.

Visualizing the Current State of the Industry

The result was a different layout that stressed more the communities of different artists who tend to collaborate with each other. Each community is colored with a different color and if there is a large central artist (or artists!), they are given a custom icon. The size of the icons are determined by the number collaborations that artist has and the thickness of the lines shows the number of collaborations between two artists. With this layout, it’s more apparent to see who each artist collaborated with and brought about the sense of communities more – while still showing the top 5! After showing this to a few people, they gave me the idea of adding written explanations for a few interactions to give more context – for example pointing out certain interactions and naming the song attributed to that interaction. 

To interpret the graph, users can find the central producers by looking for the custom icons and then looking at the lines that exit their bubble to see their reach. Chris Lake was no surprise to be central to all communities as his influence can be seen through most new artists and is one of the most impactful artists of the past decade. This graph displays this well giving him center stage (literally) with so many connections both inside and outside of his community. The dominance of Green Velvet is also apparent through the graph as he has been a staple to the industry for a much longer amount of time than some of the other artists. This version of the graph, however, does not articulate the added dimension of length of time in the industry – which would be better seen in a breakdown by time.

A Breakdown by Time Period

Lastly, I wanted to show the progression of certain artists and their communities over the past decade. To do this, I wanted to create small multiples of different time periods to show the different eras. I went back and forth between having each multiple be cumulative or distinct time periods and eventually landed on distinct time periods – 2015-2018, 2019-2021, 2022-2025. Users can track certain artists and see their growth within the industry as their collaboration and reach grows. For example in the first time period, John Summit is one of the smallest artists in a small, isolated community. By the second period, you can see where his presence in the industry grew to have his own community that is interconnected with others. By the final period, though his community is a bit smaller, you can still see how established he is and interconnected with other communities. This helps illustrate how 2020 was John Summit’s breakout year into mainstream popularity.

Future Steps

Going forward I would like to organize the small multiples to have a similar styling as the main graph. I think color coding the communities to have similar colors would be helpful for readability. Adding notes for contextualization would be helpful, but I am not sure whether I would prefer noting key interactions in each time period or following one artist – like John Summit – throughout the different time periods.

References

All data was obtained through the Spotify for Developers Web API