Posts tagged small multiples
Episode 40: How to Add Small Multiples to Your Flow Charts - Featuring Chris DeMartini

Welcome to episode 40 of Data Viz Today. How can you add more information to your flow charts? Incorporate small multiples! In this episode, we learn about Chris DeMartini's data visualization that brings together small multiples and a flow chart (an NBA bracket) to add context to the flow. Find out how he created it, and how this technique can be applied elsewhere!

Listen on Apple Podcasts, Google Play, Google PodcastsStitcher, SoundCloud & Spotify.

  • Welcome! I'm Alli Torban.

  • 00:15 - Today’s episode is all about small multiples but with a twist. If you haven’t heard the term small multiples before, it’s a term popularized by Edward Tufte, and it just means when you have a bunch of smaller charts of the same type using the same scales and axis that are bunched together showing different partitions of the dataset. It allows you to quickly compare information.

  • 00:46 - Recently I saw an interesting twist on small multiples, where Chris DeMartini incorporated small multiples into an NBA Playoffs bracket. He used the term Small Multiple Flows, which I’d never heard before, to describe when you add small multiples to a flow diagram. Even though I don’t have any interest in the NBA Playoffs, I began thinking about all the cool things you could do when you use that dead space of a flow chart or org chart or a timeline to add more detail.

  • 01:31 - Today’s featured project is called “NBA Playoffs 2018” by Chris DeMartini. Chris runs the Data Visualization team at Visa. He also works with DataBlick on consulting, non-profit and R&D projects. He has been a Tableau Zen Master since 2016.

  • 01:48 - Chris found himself a bit of a frustrated sports fan as he was following his favorite teams, like the Warriors, because when you look at sports brackets, you’re only getting the final result - who won and is advancing to the next round. But what if you want more info? Was it’s a blowout? Was it close?

  • 02:10 - He got inspired by how FiveThirtyEight does their brackets, where if you hover over a team, their bracket path to the championship changes thickness based on how likely they are to advance at each stage. Chris got the idea that he could use a similar idea where he’d encode data from the actual game into the bracket’s structure. So his goal was to have a visualization of the NBA playoffs that would give you a quick look at who is advancing but also give you an idea of the quality of the match-up.

  • 02:42 - He went through many iterations using pencil and paper to decide what information he’d encode and how he’d do it. Chris settled on a few things: First the line of the bracket was colored by the team that won the game, then the small multiple element encodes the (raw) cumulative score differential across for scoring play. So x axis is scoring play # and y axis is difference in total score as of that play. The chart is then encoded based on who has the lead on that play which carries through to lines, etc. So it’s really easy to see if there’s a lot of back and forth (data above the line, below the line) or if one team dominated. Then there were concentric circles behind each game to represent the number of wins for the team in that series. There’s a ton of info packed into this bracket! It’s a sports fan’s dream.

  • 03:42 - He said he got a lot of inspiration on how to layer all this information from vizzes around the community, but mostly from a viz by Nadieh Bremer about a Japanese Comic where she encoded a lot of data about the comic in this circular chart.

  • 04:01 - Once he figured out how he wanted to show all the info, he needed to implement it. He got the data from NBA.com and used Alteryx to get the data in a usable format. He said the data prep would’ve been the hardest part if it weren’t for Alteryx. Then he needed to execute all these encodings in Tableau. To do this, he used a technique that involves layering multiple tableau charts over each other. So in this case, he had one chart with the bracket and the colors of the lines of the brackets based on the winner, then another chart for the bars representing scoring plays, and another chart with the circles behind each game representing games won.

  • 04:45 - He has a detailed blog post about how to do this technique and another one specifically how he created this NBA viz. You can also download his Tableau workbook to reverse engineer his Tableau wizardry. You can definitely tell why he’s a Zen Master when you open it up.

  • 05:02 - So seeing Chris’ viz, really got me thinking about all the applications of this Small Multiples Flow chart. Like you could take something simple like the org chart of your team and use the lines and other elements to show extra information. Like if it’s a sales team, you can have a line chart associated with each person showing sales over time.

  • 05:25 - Or you could have a flow chart of different paths people take on your website, and add a bar chart for how many times people press the Buy button while they’re on that page. So you could see the typical path a customer could take through the website, and see that people press the Buy button quickly after landing on this one particular page, and maybe they take longer to press Buy on a different page.

  • 05:50 - It could also be a more simple flow chart, like events happening through time. Like it was recently President’s Day here in the US, and it got me thinking about Presidents and their approval rating. You could show the succession of presidents with their picture and name, and a line connecting each of them, so it’s a basic timeline of presidents, but in between each one is a small line chart showing the president’s approval rating throughout their presidency. Just this small tweak of making the timeline the main story of the graphic, and then adding data within that structure, is a really simple but interesting way to display your data.

My inspired viz - timeline of recent US Presidents and their approval ratings in between.

  • 06:40 - It might be cool to do something like this with networks too where nodes are connected with lines to show a relationship and there is another chart within those lines.

  • 07:00 - My final takeaway is that if your main story is the flow of the information, that doesn’t mean you have to leave out contextual information. You can integrate small multiples into the structure and give your audience the benefit of seeing the overall flow and extra details. So check out your flow charts and use all that space to your advantage!

  • 07:22 - Finally, I asked Chris what’s his advice to designers just starting out, and he said, “Just try it, don’t be shy, afraid or worried about what people think (which is easy to do these days, and I still do this constantly). Learn from others, be sure to site your sources and inspiration ALWAYS. Continue to evolve techniques and push viz forward in what you do.” Thank you Chris for sharing your inspiring project with us!

  • 07:47 - You can follow him on twitter @demartsc. And check out his great blog posts on the datablick blog.