Are You Up To The Challenge? My Creative Process for Data Viz Challenges

Have you heard of #MakeoverMonday or #WorkoutWednesday?  I’m sure many of you have, but if not, they are two of the many free and publicly available challenges offered by the Data Visualization community.

Some challenges are offered weekly while others are monthly.  Some focus on creation of new visualizations while others focus on challenging you to recreate existing ones – with the goal of teaching new visualization techniques.

For me, there are three main reasons why I have become a regular participant in these challenges over the last year and a half:

1. Practice

As with any art form, to get better at data visualization you need to practice. The more that you create, the more scenarios that you encounter, the more functionality that you’ll be able to understand and use, and the better your skills will become.  You can never stop learning or get enough practice with data visualization.  Even Tableau Zen Masters are constantly learning and working to expand their knowledge.

2. Accountability

While it’s easy for me to set goals for myself, if I don’t have a means of staying accountable to those goals, I have a much harder time of staying on track. Creating these challenge visualizations, publishing them to Tableau Public, and then posting them on social media has been a great way for me to stay accountable to my progress.

3. Building a Portfolio

Along with accountability, as I continue to publish my work on Tableau Public, I’ve amassed a nice portfolio of non-job related data visualizations. This has been especially useful when I’ve applied for new positions in the past.  You usually cannot show potential employers visualizations you have done at work, but if you have worked on things on your own time, and have them published on Tableau Public, you have a portfolio of your work in reverse chronological order.

My Creative Process

I just recently participated in #ProjectHealthViz, a monthly challenge by Lindsay Betzendahl focused on publicly available health data.  Health data is fun to visualize as there are many different stories to be found within the data and many different ways of presenting them.

This month’s dataset was from the CDC and focused on disease outbreaks across the US.  The data was in an excel file with over 400 thousand rows.  Each row contained information on a single outbreak.

Let’s walk through how I got from the raw data to my final viz.


Norovirus Outbreaks!/vizhome/ProjectHealthVizNovemberNorovirus/NorovirusOutbreaks

Process, Not Rules

While each challenge is different, when I’m creating something new there is a general process I follow.  By no means are any of the following points meant to be seen as rules (very few of those exist in data visualization).  Rather my intent in sharing these guidelines is to provide a spark to those who are just getting started with these data viz challenges and provide a possible structure for doing so.

Reviewing Data Structure

The first thing I do when presented with a new dataset is to look at which data fields are available.  In this data from the CDC, there were:

  • date fields for when an outbreak occurred
  • state location information
  • an Etiology field that listed the name of the disease
  • some descriptor fields (like mode of transmission and outbreak setting)
  • several measures including the number of illnesses, hospitalizations, and deaths.

When reviewing the structure of the data, you want to get a general understanding of what your main dimensions and measures will be for your visualization.

Exploration and Choosing a Topic

This CDC dataset is a little on the heavier side as far as content goes but it’s rich with stories to be told.  To decide which story I wanted to tell, I needed to explore the data.  While I could have done this in Excel, I like to bring the data into Tableau to start exploring.  I start by reviewing overall statistics with the goal of finding a topic that stands out in the data.

For this dataset, that included the following:

Number of Outbreaks Per Year

Outbreaks per Year


Total Outbreaks by Location

Outbreaks by Location


Outbreaks by Location 2


Number of Outbreaks by Disease Type

Outbreaks by Type

It was clear from my exploration that Norovirus had the most outbreaks of all the diseases in the dataset.  I decided at this point to use Norovirus as the focus for my viz.  While I don’t always create visualizations based on the highest or lowest numbers in a dataset, once I have found a good topic I like to start creating my viz in Tableau.

Although you can and should take as long as you’d like when creating your own visualizations, I try to keep some structure around the amount of time I spend on any single viz.  Some take longer than others, but I have a family and friends that I like to spend time with so I make sure my viz time does not take away too much from my personal time.

Telling a Story

Sometimes you have a dataset and you know exactly the story you want to tell right away.  In those cases, you may start by creating the overall design of your dashboard and filling it in with your data later.

I tend to start with the data before I create a dashboard.  I create several different charts, and iterations on those charts, before I put them all together.  For the #ProjectHealthViz challenge, I created several different views.  However, not all of them made the final version of my dashboard.

It’s up to you how many charts you want to create before combining them into a dashboard, but don’t feel like everything you create needs to be included.  Sometimes your best ideas don’t come till after several iterations or a few times stepping away from your computer and coming back to it.  More importantly, even if you really like one of the several charts that you have created, if it doesn’t fit with your overall story, you do not need to include it in your final viz.

Putting It All Together

Once I have a few different views to choose from, I start putting them together.  My dashboards usually are not the same shape or size, and I often resize my dashboards as I’m working on them.

The one actual rule I stick to though is using fixed dashboards.  You never know what size screen your viewers will have and if you do not fix the size of your dashboard, it may stretch or shrink in ways that make it unreadable.

Alignment and Chart Selection

There are many different ways to layout your dashboard.  I often like to make a rough sketch with a pencil and paper of what I want my dashboard to look like.  This helps me visualize what I want to do before I start creating in Tableau.

For my ProjectHealthViz I decided that I wanted to make it look like a long and narrow infographic – something where the story builds as you go from top to bottom.  While I had several charts to choose from, I decided on these three to present the data:

  • Bar Chart
  • Hex Map
  • BANs (Big Ass Numbers)

I’ve used bar charts and BANs many times in the past.  They’re both great ways to highlight simple information.  The bar chart I created clearly shows that Norovirus is the top cause of outbreaks in the US.  The BANs at the bottom deliver raw numbers for how many have been affected by Norovirus.

As for the hex map, while I could have used a traditional map instead, there are two reasons I chose the hexagon style.

  1. It makes it easier to see the number of outbreaks per state when each state is the same size.
  2. I’ve always wanted to create a hexagon map and hadn’t done so before.

While some charts are more effective than others for different scenarios, sometimes stretching your ability by trying something new is fun too.  And data viz should be fun!

Font and Color

I’m no expert on fonts.  However, Tableau Public has a limited number of fonts it can render.  Since the options are limited, it’s not too difficult to choose a font.

Jennifer VonHagel put together this great resource on which fonts are supported by Tableau Public and how they look:

I personally like the look of the Georgia font but sometimes stick with Tableau Regular (and use Tableau Bold or Tableau Semibold for bold characters).  For my ProjectHealthViz, I stuck with Tableau Regular and Tableau Semibold.  The bold function does not work well with Tableau Regular so I used Tableau Semibold for making certain text and numbers stand out.

When it comes to color selection, I try not to be too flashy.  You don’t want extra color taking away from what you are presenting with your data.  You also don’t want to use color that doesn’t have meaning.  Color can be a great asset in your viz, but it is also very easy to overuse.  My recommendation is that less color (and a fewer number of overall colors) tends to work better for visualizing data. In my viz, I used red to highlight the outbreaks while keeping everything else black or white.

Custom Fonts in Tableau

For my title at the top, to give it a little something more than the standard font look, I turned to PowerPoint.  You can create a title in PowerPoint using any font available, save it as an image, and import it into your viz as an image.  This way it looks like you’re using custom fonts, even though they’re actually image files.

At the end of last year, Zen Master Ken Flerlage tweeted about a tool called that lets you type in a word or phrase and see how it looks in all of the fonts installed on your machine.

I’ve used on several of my visualizations. It’s an easy way to find a font for a custom title, especially if you do not want to go through each font in PowerPoint one at a time.


The footer is where I place information to tag my viz as my own.  I put the name of the challenge I’m participating in and my Twitter handle.

This is also where two very important pieces of information should go:

  1. Link To Your Datasource:  If the data is publicly available, make sure to cite where it came from.
  2. Attribution: If you were inspired by someone else, or took a similar approach to a visualization as someone else, make sure to note that in the footer. Attribution in data visualization is as important as attribution in a book or research paper.

Be Deliberate About Difference

Once your viz is done, make sure to review it.  Check for consistency in both your font usage and your colors.  If you decide to make something a different font or color or shape than other items in your viz, do it intentionally and know that you did it. Do not do it because you overlooked it.  Be deliberate about difference in your viz.

Check your tooltips too!  They’re often overlooked and can be a useful feature for including additional information.

Pro tip: It never hurts to have someone else review your viz.  The Data Viz community can be a great resource for this, but sometimes it helps to have someone from outside the field review it as well.  If they have trouble interpreting your viz or understanding your data, you may need to make it clearer.

Get Started!

Sometimes getting started can be the hardest part.  Regardless of the process you use, the most important part of getting started with these challenges is to create something.  Import your data into Tableau and start making charts.

I’d say the second most important part is sharing your finished viz online.  For me, sharing my work online has been the best way to stay accountable to my progress and to elicit feedback from the community.

Here’s a list of challenges that I have had the pleasure of participating in (in no particular order):

 I challenge you all to participate in more challenges.  And if you ever would like any feedback on your visualizations, I’m more than happy to help!

Searching for My Inner Artist

Inspiration at Tableau Conference 2018

TC18 left me inspired, empowered, wanting to do everything, and also a bit tired.  Like many others I wanted to rush to complete a blog as soon as I left New Orleans.  However, as often happens, we start out with great intentions but life intervenes.  I started writing in fits and spurts, finding it difficult to get an opportunity to sit down and get my thoughts out.  At the very least though, I knew what my blog topic would be.

On the last day of the conference, I had the great fortune of attending Giorgia Lupi’s talk on Data Humanism.  A month or so before the conference I had seen a wave of people in the Tableau Community tweet about her Observe, Collect, Draw book and start creating wonderful artistic imagery, outside the realm of our normal day to day data viz.  They weren’t just creating charts or reports – they were exploring their own creative ability.

My Data Viz Background

I have to admit that the artistic aspect of data visualization does not come naturally to me, and it’s the piece that I want to improve upon the most.  I’ve never been artistic.  My ability to draw by hand didn’t progress much beyond elementary school.  I can be creative poetically and occasionally musically but visual creativity has always been hard for me.

I did not come to Tableau for its artistic capabilities.  I began my career in business analysis type positions where I used a ton of Excel and looked at a lot of numbers.  Most of my professional Tableau use has been more on the data analysis side.

Ask a question > find an answer > use the visualization as a means of finding the answer.

In the data viz community, especially in professional environments, visualizations that don’t provide quick and obvious impact are sometimes scoffed at and considered “data viz as art”.

Seriously, You Should Watch This

While I had begun to appreciate the artistic side of data visualization from connecting to the Tableau community on Twitter, Giorgia’s talk at TC18 was the type of experience that would change the mind of any data viz purist.

If you want an example of the experience, read about her work with Kaki King called “Bruises: The Data We Don’t See” in this Medium article (link) then watch the video they created here:

I’ve never been that moved by data visualization before.  Bar and line charts can’t convey that type of emotion (don’t @ me).  The feeling that her worked evoked in me was encapsulated in her definition of data humanism – “using data to become more human instead of more efficient.“   That sentiment had not crossed my mind before that point.

Additionally, her Dear Data Project with Stefanie Posavec showed me that inspiration for artistic data viz can come from anywhere and does not need to follow standard conventions. My top takeaway quote from her session was, “it’s much more interesting to see how we can expand definitions of data viz instead of constricting them”.

Find Inspiration Everywhere

As I was procrastinating from writing this blog by perusing Twitter, I found that Tableau Zen Master Neil Richards had already included a great tribute to Giorgia’s talk at TC in his post TC recap.  While I highly recommend reading it (link), it also took me to the recording of his TC talk “Design Driven Data”.  Unfortunately, I had been unable to see it live, but I’m so glad I watched it later.

In it, Neil discusses his own non-artistic background and how he’s been able to use Tableau to explore his creative ability.  Instead of the typical approach of data driving the design of a visualization, he implores his audience to find data that fits an original design that you create.  Doing so may lead to better results, and possibly have more impact than a traditional data driven approach.  His response to those that say that “data art isn’t proper visualization”?  “Good data art still visualizes the data.”

After watching both Giorgia and Neil espouse the beauty of design and captivation in data visualization, and in admiration of Neil’s ability to find inspiration anywhere, I was ready to create my own piece of design driven data art.


While I was searching for inspiration, I noticed a dreamcatcher at the Airbnb in Austin I was staying in the week after TC.  I appreciated it’s design and colors, and most importantly I believed that I could create something like it in PowerPoint.  I used Kevin Flerlage’s method of creating polygon-like vizzes without polygons (link) to get to work.

Before I had any data, I created a Dreamcatcher design with some random colors in PowerPoint.  Each piece of the dreamcatcher is an individual shape.  I saved individual png images of each of the 32 shapes so that I could use them later in Tableau.

Dreamcatcher Shape

Embracing the design driven data approach, I now had my design but did not yet have data to go with it.  I focused on the 32 shape image files I now had.  The only two things I could think of that have the number 32 are the number of teams in the National Football League and the number of teams in the World Cup.  As I’m a big fan of the NFL and know much more about American Football than Everywhere Else Football, I decided to use NFL data.

For the color, I utilized this fantastic list of 100 different color palettes from Canva (link).  It was both a blessing and a curse as I tried out many, many, different color combinations before deciding on one I liked.

Now that I had the design part down, my traditional data viz and analysis background kicked in.  I needed to find some data that worked with my design.

Data Behind The Design

While Kevin’s blog shows a method for using size to differentiate each shape, since I have 32 shapes (each representing one of the 32 NFL teams) and they are in very specific positions, making each a different size would cause them to overlap each other.  Instead, I decided to find some binary, yes/no type data, where a shape would appear if its value in my dataset was “yes” and would disappear if its value was “no”.

There are many sites that have NFL statistics.  On, I saw they had data on each team’s Passing Play Ratio – the percentage of a team’s plays that are passes versus those that are runs.  I decided to use this for my binary data.  More than or equal to 50% passing equals “yes”, below 50% passing equals “no”.

To decide how to position each team’s shape, I started at the very top.  The shape at position 12 o’clock is the top ranked team for that year.  The teams then decrease in rank clockwise around the circle.

Dreamcatcher - original

Over a period of 16 years (starting at the top left and working to the bottom right, if a team passed 50% or more of the time, their shape appeared on that season’s dreamcatcher.  If they passed less than 50%, their shape is hidden.  The viz shows that passing is becoming more common (less shapes disappearing over time).  In fact, over the last 4 seasons, only the 2018 Seattle Seahawks have run more than they’ve passed – and the season’s not yet over.  I’ve included viz in tooltips to show more traditional data on each team’s Passing Play Ratio.

Honestly, the hardest part of creating this viz, more so than creating the shapes in PowerPoint, or gathering my data, was creating the legend to go along with it.  An often overlooked and under-appreciated aspect of artistic data representation is the legend that explains what you’re seeing.  This is the piece that differentiates data-based art from non-data art.  Since my viz is built off of NFL data, I need to inform the viewers of what they are seeing and how to read it.  I’ve also leveraged viz in tooltips in my legend.  My intention in doing so was to save space and maintain the height of the original Dreamcatcher viz.

Dreamcatcher with legend

And here is my final piece (Tableau Public link).  This is not something I would have considered creating before TC.  Having done it though, I’m eager to see what other types of artistic data viz I can create.  Not everything I create is or ever will be the most artistic or most visually appealing, but at this point I feel like my artistic exploration is more important than the outcome.