On a daily basis, my job involves visualising data, this may be a line chart to accurately monitor sales, a bar chart for daily CRM reporting, or even a trusty pivot table to deliver business insight. But what I don’t usually do when developing a visualisation is think “Oh, well that would look great hanging up on my living room wall”
It was only recently I stumbled across a gentleman named “Kirrell Benzi” on a subreddit astutely named r/dataisbeautiful. He had created a piece called “The Dark Side and the Light” which you can see for yourself, among his other work, at https://www.kirellbenzi.com/art/dark-side-light. It’s a stunning visualisation which assigns every character in the Star Wars universe (which is over 20,000 btw) a node, these nodes are then connected together and coloured depending on their relationship, resulting in this explosion of colour. In total Kirrell used 20,010 nodes with 66,425 edges or connections.
Drawing on this newly found inspiration, I decided to try and create my own Data Art Piece, but where to start? I wanted to create something personal to myself so I thought I would look at the connections in my life, friends, family and the like.
The only way of quantifying the number of people in my life and connections without digging out my address book from the ’90s (which, let’s be honest, will be a little outdated now) was by using a more modern social approach, Facebook.
The beauty of retrieving the data of my “friends” from Facebook is that we should also be able to find the data which would distinguish the connections between people: mutual friends and so on. A little bit of research, Googling brought me to a Google Chrome extension called “Lost Circles”, a tool that scrapes your profile and pulls back all of your friends and connections! Perfect for a lazy wrangler like myself.
Once Lost Circles had done its thing, I was given two options: an export to JSON (which is no one’s friend) or a GraphML file. I had never come across a GraphML file before, so I did a little more research, Googling about where these files can be used. I was introduced to a tool called Gephi, (https://gephi.org) an open graph visualisation platform which upon inspection I immediately realised was the tool for me.
Importing my GraphML file into Gephi I was automatically greeted with all of my friends being represented as Nodes, and the connections between each represented by the lines. There were 897 nodes in total and 9568 connections, which is most certainly a lot smaller than the example Kirrell had put together but definitely something I could start to work with.
I began to work on the clustering of my friends’ groups; making each cluster or circle of friends distinguishable by using the Fruchterman-Reignold force-directed layout algorithm.
This successfully grouped my friends into clusters, and by hovering over the nodes I could distinguish where those groups were formed. It was also interesting to see how my family and friends were all connected and found a good amount of people that had connections across clusters that I didn’t know about! The force direction also turns all of the nodes into a large spherical shape which aesthetically looks really good as a base for my “art”. The only remaining steps with regards to preparing the data were to remove out any outliers, for example, people I may have added in a Nightclub or at a business meeting, anyone that really isn’t connected to anyone else.
Below you can see the main clusters that were formed:
Now, to channel my inner Neil Buchanan, I would need to colour the nodes and lines accordingly to the clusters and how they relate. This could be achieved within Gephi by passing through another algorithm (Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte, Etienne Lefebvre, Fast unfolding of communities in large networks, in Journal of Statistical Mechanics: Theory and Experiment 2008 (10), P1000), this groups the nodes into modules which can then be coloured accordingly. To choose my colours I went to a great website called coolors.co which helps you pick a palate which works together, I decided to go for an 80’s theme with lots of pink to reflect my obsession with anything synthwave! I then turned down the size of the nodes to make the image look almost neural in aesthetic.
Once finished in Gephi I could export the visualisation, which can either be a PNG, SVG or PDF which is handy. I popped it into Adobe Photoshop for some final magic touches and here is the result.
