tapestry conference

At the end of November, I had the pleasure of attending the Tapestry Conference in Miami. I don’t attend a ton of conferences and this is actually the only one that exists where I’ve (two years in row!) been present for every single session (both physically and consciously) and found something useful or inspiring in each one. If you’re reading these words with slight envy for not having been there—I can’t recreate the great break-time chit chat with attendees, but I can share the presentations (huge thanks to organizers for making these available): here are the videos.

In particular, I’d recommend the keynotes. Mona Chalabi opened the conference with an entertaining session discussing a number of her hand-drawn graphs (a quick scroll through her Instagram will give you a sense of your work if you aren’t familiar; unfortunately her talk isn’t being shared). She described wanting to feel something about the data and marrying the subject and the visualization so that if you see the visual, even without labels someone can get some sense of what it is about. She also worked in good reminders on significant digits (too many conveys false sense of precision), designing with visual impairments in mind (using alt text or sound, like in this work), and how important the simple question “do you get it?” posed to people unfamiliar with your topic can help point out issues or help you to identify improvements.

Matt Kay’s keynote on Uncertainty (“A Biased Tour of the Uncertainty Visualization Zoo”) was fantastic—he made the point that it isn’t necessarily true that people aren’t good at understanding uncertainty (a claim often made) and that there are intuitive ways to communicate uncertainty that we should be using. I like the onus this puts on the designer of the information. Matt illustrated several specific methods—icon arrays, quantile dot plots, and animating—for better communicating uncertainty. I also learned a new term: subitizing, which describes how we can see a small number of something, for example three circles, and we recognize (without counting) that there are three. This is both useful to be aware of when designing graphs and also simply a word that I will enjoy adding to my vocabulary.

Elijah Meeks delivered the closing keynote on the “Third Wave of Data Visualization.” He describes wave one as Tufte-inspired with the goal of clarity and the second wave of systems following Wilkinson’s The Grammar of Graphics, leading into the third wave of today. Rather than tell you more about it, I encourage you to listen to Elijah tell you about it directly (plus more!) in Episode 12 of the storytelling with data podcast.

In addition to the keynotes, there were eight short stories (roughly 15 minutes each, standout ones for me were Jonni Walker’s and Alex Wein’s) and a number of short talks (about 5 minutes each). You can hear Jon Schwabish and me chat about more of the sessions in our Tapestry roundup. I highly recommend watching the videos of the Tapestry presentations.

Big thanks to organizers, speakers, and attendees for combining to make this an awesome event (and extra thanks to the organizers for recording and making the content widely available!).

let's visualize the holidays!

It’s December! I’m not entirely sure how that happened, but as you know, a new month means a new #SWDchallenge. Last month, I think I scared some people off when we ventured outside of data visualization land with a sticky note focused exercise (challenge | recap post). I remain a strong believer in this sort of low tech planning, so don’t be surprised if we revisit something similar in the future. But this month, we’re back to graphs with an open-ended holiday challenge.

‘Tis the season to be merry (irrespective of which holiday you celebrate), so let’s combine that festive cheer with something near and dear to us all: data visualization! This month, your challenge is to visualize data related to the holidays. I welcome you to create anything you’d like that combines data and holiday (and any holiday is fine, though given the timing I’m guessing we’ll see a lot of winter-related ones!). Data exists (or could be compiled) on all sorts of holiday-related things: where various Thanksgiving food items come from, the cost of Christmas, whether certain holidays were historically snowy in your city of interest, or popular holiday music, for example. I’ve started a list here, but it is definitely incomplete—if you encounter any amazing sources of holiday-related data, or create any that you’d like to share please add them!

In case it provides inspiration, here are a couple of my historical holiday data viz creations, with links to accompanying blog posts:

Visualizing the cost of Christmas

Graphing holiday wishes

My challenge to you: find some holiday-related data of interest and graph it! You can look to past challenges for ideas related to trying a certain type of graph or practicing a specific strategy (takeaway title, intentional use of color, sticky note storyboarding). I welcome you to use this challenge to simply practice effective data visualization or try something new. (As a bit of foreshadowing, I’ll ask us to all try something new in the new year!)

DEADLINE: Monday December 10th by midnight PST. Full submission details follow—be sure to email it to us, taking note of specifics below, for inclusion in recap post! You're also welcome to share at any point on social media using #SWDchallenge.

SUBMISSION INSTRUCTIONS:

• Make it. Identify your data and create your visual with your tool of choice. If you need help finding data, check out this list of publicly available data sources or here’s a special one focused on the holidays for this month’s challenge. You're also welcome to use a real work example if you'd like, just please don't share anything confidential.

• Share it. Email your entry to SWDchallenge@storytellingwithdata.com by the deadline. Attach your image as a .PNG. Put any commentary you’d like included in the follow up post in the body of the email (e.g. what tool you used, any notes on your methods or thought process you’d like to share); if there’s a social media profile or blog/site you’d like mentioned, please embed the links directly in your commentary (e.g. Blog | Twitter). If you’re going to write more than a paragraph or so, I encourage you to post it externally and provide a link or summary for inclusion. Feel free to also share on social media at any point using #SWDchallenge.

• The fine print. We reserve the right to post and potentially reuse examples shared.

We look forward to seeing what you come up with! Stay tuned for a cheery recap post sharing back the submissions received later this month. In the meantime, check out the #SWDchallenge page for past challenge details and recaps.

recommended reading: #MakeoverMonday

I arrived home from a trip a few weeks back to a fun package containing a thematically wrapped copy of MakeoverMonday: Improving How We Visualize and Analyze Data, One Chart at a Time. I had a chance to review and provide feedback on an early draft chapter of this book, cowritten by Andy Kriebel and Eva Murray, which was published last month.

Just yesterday, I spoke with Andy and Eva about #MakeoverMonday: the project, the book, their writing process, and more, which you can listen to in the latest episode of the SWD podcast.

At more than 400 pages, #MakeoverMonday is a hefty book and contains hundreds (300, to be exact!) of visuals created by more than 50 people. The book was born out of the weekly #MakeoverMonday project that Andy has been doing for a number years (joined by Eva in 2017). Each week, they select a graph and invite people to critique and improve it. People share their creations (see Twitter: #MakeoverMonday). Andy and Eva also host weekly live viz reviews, where they provide feedback to a subset who volunteer their work, with the expectation that the creators each iterate based on that feedback and re-share. More info can be found at makeovermonday.co.uk. Andy and Eva curated examples from this weekly project as the basis for their book.

Given the sheer number and variety of visuals in #MakeoverMonday, it’s great to flip through to see examples of things done well—and some cases of things gone wrong. Many learnings and lessons are recounted, from the habits of a good data analyst to working with incomplete data to trying new things. In particular, I appreciated Chapter 8, “Iterate to Improve,” which focuses on the important role of giving and receiving feedback and illustrates through a number of before and after examples how minor changes can have major impact on the overall effectiveness of a visual. If you analyze and visualize data—or would like to—this is a book you will likely want to have on your shelf.

As a bonus, data from the weekly challenges that feed the visuals profiled in the book are all available for download. This is a great resource if you ever find yourself in need of data to visualize or if you want to recreate or further improve upon any of the examples in the book.

#MakeoverMonday can be purchased on Amazon. Listen to my chat with Andy and Eva in Episode 11 of the SWD podcast.

I used an area graph!

I don’t use a lot of area graphs. In fact, they weren’t even included in the visuals I use most that were covered in storytelling with data: a data visualization guide for business professionals (tangentially related: all the data and visuals from the book are now available to download). That means I didn’t use any area graphs for a whole year!

I’ve mainly avoided area graphs for a couple of reasons. They take up a lot of ink, simply due to the typical amount of filled in color and limited white space. This can make them feel a bit heavy, and may mean there isn’t space left for annotations or labels the way that you often have with a line graph. They are prone to some shared issues with stacked bars: if anything interesting is happening further up the stack, this can be hard to see because it’s stacked on other things that are also changing in size from point to point (the connected lines and area, however, seem to help make up for some of this, compared to the disjointed view you get with bars given the white space in between them). Finally, I often find it unclear whether the series in the stack are literally stacked on top of each other (each graphing from the bottom of the given segment to the top), or if they overlap (each graphed from x-axis upwards).

All of that said, I’ve found myself using a couple area graphs lately, and so thought I’d share one example with you. This is from a recent client makeover, where I iterated through a number of different views in order to both get a better understanding of the main point I wanted to make as well as identify an effective visual that would help me do so. The team in this case was plotting the capital budget outlook over time across five categories of spend. The original graph looked similar to the following:

There are a number of things that are not ideal about this initial view. One of the first things I notice is the y-axis and the (000’s) label at the top. A thousand thousands is a million, so I’ll likely drop some zeros and change the scale to millions of dollars. When I look at the x-axis, it starts to become clear that some of this data has already happened, while other dates are in the future (reflecting some sort of forecast or plan). I’ll want to make this visually clear. I can do this in a couple of ways: I can add words, and/or I can make the actual data points for the future dates somehow visually distinct (for example, lighter color, or lower intensity). The following graph incorporates these changes:

Even after I’ve made these amendments, however, a larger issue remains: I have no idea what I’m supposed to do with this data!

I’ve encountered a number of situations similar to this lately, where an analyst or an analytical team views their role first and foremost to inform. The sentiment seems to be something like, “I’ll put the data out there, and my audience will know what to to with it.” This is a far too passive view of the analyst role, in my opinion. If you are analyzing data, you likely know it better than anyone else. This means you are in a unique position to drive value based on that data. Don’t simply show data, turn it into information that your audience can do something with and act upon!

Let’s spend some time looking at the segments being graphed and considering what they represent. We see the budget for three major projects decreasing markedly from 2018 to 2019, and then decreasing slowly over time:

There is also decreasing trend for other existing projects, which look to be at zero by 2021:

The budget for new projects increases a bit over time:

That leaves two final pieces: existing allowance and proposed increase. Hm. There might be something interesting here. But before we jump to that, let’s get a little clearer on what we’re looking at. While the three data series up to this point have been stacked one on top of the other, that changes when we get to the existing allowance. The existing allowance is actually the first three data series combined, plus a small amount in excess of that—we only see this latter bit graphed separately:

The final series, proposed allowance, is all of this together, plus the incremental part that is shown explicitly with the top blue series currently:

So we aren’t even graphing this data quite right currently.

Let’s unstack the bars and focus on the total amounts for these final two data series (I’ll also use this as an opportunity to adopt a color scheme that is more fitting with the branding of the organization, the logo of which is bright blue and dark blue-grey):

One of the most interesting parts of the data here is how the proposed allowance compares to the existing allowance, so let’s actually get rid of everything else:

It will be easier to focus on the difference between the bars in the preceding view if we turn them into lines:

The really interesting thing here is the gap between the lines, so let’s focus on that:

At this point, I decided to play with the design a bit and put words on the graph to make what we are looking at clear:

Next, if we want to pull the breakdown of the original first three data series back in, I can do that with a stacked area graph:

In this final view, focus is meant to be on the gap between the existing and proposed allowance, which we can see amounts to $195M over the period from 2019 to 2021. I can still see the 3 major projects and other existing projects decreasing over time, as well as the slight increase in new projects, so that detail from the original view is preserved but kept from being distracting by graphing it all in the same (relatively muted) color and labeling directly. I focus on the thing that is different: the gap in bright blue. There are words, both in the title and in the right hand margin that help me understand what I’m looking at and what action needs to be taken. Looks like we need some additional budget!

In case it’s useful, you can download the Excel file with these graphs.