When I was new to #dataviz, seeing only finalised projects by seasoned practitioners used to puzzle me: I always have thousands of questions on the process: how did you get there?
That’s why I document most of my projects, explaining the reasoning, design choices, technical hurdles etc.

This is the third time I enter Ironviz, Tableau’s dataviz competition, after an entry on Buffy The Vampire Slayer in July 2021 that got to the Top 10 Qualifiers, and an revisitation of a Hamilton viz I described on this blog here and here.

This year, the theme was games. As I’m not at all into sports games and have no idea around a video game viz, I went with a viz related to the board game I play (online) the most.

#1 The concept and the audience

This piece is supposed to a fun introduction to the game, not a complex statistical exploration of how to win – some people have done it more thoroughly than I did.

Why? Because my intention was to keep a broad audience, and to be accessible to people who have never played.

You’ll notice that I don’t explain all the rules and introduce the complexity of the game sequentially.

My whole point is that the relative complexity of the game is why I love it so much… despite the fact I’m not really good at it.

#2 Getting the data

BoardGameArena doesn’t allow webscraping. After checking with the website point of contact, they allowed me to pull my data manually that I then had to clean it in Excel.

Note: I’m quite glad I worked on my Excel skills lately, because it made the data cleaning much faster. No matter what the data professionals will tell you, (advanced) Excel remains still the most transferrable skill you could ever learn. So, yes, don’t neglect to learn Excel (advanced formulas, Power Query), for instance from Oz du Soleil, his LinkedIn courses are hands-on, fun and engaging or you can check his YouTube channel.

#3 Design and colour

I wanted the colour palette to reflect the card colours from the boardgame, hence the green representing science, red for the military structures etc.

I spent a bit of time looking at colour palettes and then decided… that the Tableau one was good enough: Tableau’s designers spent hours honing the palettes, and it works very well for my topic. That’s not the first time I got with this shortcut. No shame, just steal like an artist!

For the fonts, the body of the text needed a web safe font as per Judit’s blogposts on Tableau and fonts rendering.

For the second display font, I wanted maximum contrast compared to Trebuchet: I went with a handwriting one, Sue Ellen Francisco. I like handwriting fonts because to me, they signify or go well with quantified self datavizzes, such as my latest Depeche Mode one.

I had a couple of options in mind and tested it with some family members and went with the one that was described as the most dyslexic friendly.

#4 The charts

For me, Ironviz is an opportunity to challenge myself and try something new every time.

It took me AGES to find charts I was happy with.

I toyed with this visual from Ellen Blackburn for a long time before but in the end went with a dumbbells / slope chart inspired by Ryan Soares’ Premier League Summer 2020 Transfer Spending that Ken Flerlage shared in his blog Proportion Plots in Tableau.

As for the last chart, ideally, I think I could have gone with a Sankey or dendrogram so that you could see the 3 ages in hierarchy and the links between the cards.

But I find uses cases for chord diagrams so rare, that I thought I would be a great opportunity to try.

First thing I did was test it in Gephi first.

Why? Because

1. I know the tool and I wanted to have a first look at the possible output
2. Even if I hadn’t managed to create the curved lines, I knew I could have created straight ones, as I did in my Buffy viz.

But this time, I went further with curved lines!

I followed Brian’s Moore tutorials Fun With Curves in Tableau Part 1: Circles and Part 2: Bezier Curves.

• If you download the workbook, you’ll notice that I have ended up hard coding the localisation of the circles (I call them nodes with the Ken Flerlage method explained here but I could have just sticked to Brian’s tutorials in the end (see later, section ‘The one thing I wish I had done differently’).
• For the curved lines, I reached out to Brian because the initial calculations created this chart and I thought it was a bit messy

I reached out to Brian and asked if there were ways to edit the “bounciness” of the curved lines. He kindly explored the topic a bit more and I have the honour to have tested the method 3 he describes in this blog Fun With Curves in Tableau Part 2+: Controlling Bezier Curves.

#5 The bonus

I always feel a bit guilty when I publish a quantified self dataviz.

There’s a part of me that wonders “why should my audience care? Why are you making this about you”.

My biggest inspiration to embrace quantified self is Michelle Frayman: yes, you can write about yourself without being self centered. It took a generous person like Michelle to teach me that it’s ok to make your viz personal.

That being said, in a quantified self dataviz, I will always look for the opportunity to add a “what about me” for my reader.

For instance, when I looked at the band Depeche Mode, and why I love some albums more than others, I also gave an opportunity for my readers to explore more data so that they can craft their own story.

In the case of this 7 Wonders viz, I have added in a hidden container all the parameters to allow you to enter your BoardGameArena data instead of looking into mine! Some of the text analysis then changes to adapt to your results.

#6 The one thing I wish I had done differently

I wish I had tested earlier the animations in Tableau public!

Because the viz is so complex (26 charts, including the intensive curved lines), the animations won’t load.

Per se, I don’t mind about animations. But it’s the fact that the cursor doesn’t change to a hand when an element is clickable that bothers me the most.

I attempted to find a solution, by using a less intensive way of laying out the circles (which I call “hard coding the nodes”) and I reduced Brian’s Bezier model from 50 to 30 but no luck.

In the end, I ran out of time.

What I could have done differently: instead of a one pager viz, if I had broken it down into several dashboards, probably the animations would have worked on the rest of the viz, if not on the chord chart!

However, I am very happy with my viz overall.

It’s playful, light-hearted, includes an Easter egg and it reflects the fun I had thinking about the topic.

And a massive thanks to everyone who helped me: Michelle Frayman, Zak Geis, Nicole Flassen and Klaus Schulte for their feedback during the Ironviz feedback sessions and VizOfficeHours, and Brian Moore for the calculations on the last chart!