Have you ever found yourself staring at a blank Tableau canvas, unable to start a viz? I have. From vizzer’s block to being paralyzed by choice, I know that gray canvas well.

This week, I found myself in a familiar situation. I had a great food-survey data set, bursting with stories waiting to be told but where should I start(1)? Thankfully, I have a few tricks to help overcome the blank-canvas blues.

1. Start drawing

Drawing/sketching/doodling (whatever you want to call it) is one of the best forms of brainstorming I know. The sketches don’t have to be pretty or even legible. Putting pen to paper is my best way to kickstart the creative process.

And I'm not alone in this camp. From academic papers and news articles to professional visual communicators like Catherine Madden, everyone is drawing.

2. Follow inspiring visualizers

One of the best ways to be inspired is to surround yourself with inspiring people and vizzes. For me, this means following data vizzers and data journalists on Twitter, reading data-viz books, and taking notes (or screenshots) of vizzes I like.

This tactic paid off. I came across this visualization by the Washington Post, and immediately saw how I could make a similar viz to tell my food-survey story.

Don’t know who to start following? Check out Andy Cotgreave’s data visualization Twitter list and follow Viz of The Day on @tableaupublic (or get it delivered to your inbox).

3. Have a checklist to clean and analyze your data

It might sound counterproductive to get your creative viz juices flowing by following a checklist, but structure can help overcome vizzer’s block. My checklist is divided into two parts, data preparation, and data exploration.

Data preparation

As dull as it sounds, physically looking at your data helps you understand the data set’s possibilities and limitations. Here are some of the things I look at in a data set:

  • What fields does the data set contain (and not contain)?
  • What kind of data does each field contain?
  • How is the data structured and formatted?
  • What are the minimum and maximum values in each field?
  • Do any fields contain null values?

By going through this list, I could see that my food-survey data set had multiple levels of details in it—some food had up to four sub-categories while others only had two. That meant it would be hard to meaningfully compare two food items unless I knew that they were both at their lowest sub-category.

Data analysis

I think of analyzing a data set as a way of interviewing it. If I’m stuck staring at a blank Tableau canvas, I can fall back on asking traditional interview-style questions of my data:

  • Who, what, how, why, when, and where? Go through each field and see how you can apply one of these questions to it.
  • Embrace your inner-child and ask, "Why? Why? Why?" of your data.

Here's one line of questions I asked of my data:

  • Question: Which category had the sharpest consumption decline when compared to 1974?
    Answer: Sugar and preserves.
    “But wait, isn’t this the opposite of all the articles I have read saying our sugar consumption is at an all-time high?”
  • Question: When did the decline start?
    Answer: In 1975.
    “This is really different from what I thought. Why is it happening?”
  • Question: What food stuffs does this category contain?
    Answer: This category includes raw sugar (think: a bag of brown sugar).
    “Ah, maybe people are buying fewer bags of sugar from the supermarkets. But are we consuming more sugar in other forms?”
  • Question: Are other sugary categories like soft drinks increasing?
    Answer: They are. Both soft drinks and confectionery are increasing.

As you can see, interviewing your data gives a structured way to beat the blank-canvas blues.

4. Remember: Building a viz isn't a one-way-street

Sometimes I find myself unable to start vizzing because I’m worried that the final viz won’t show the deepest insight, or won’t be the best way to tell the story. Remember that there are an infinite number of ways to approach a viz, that there isn’t only one best story to tell with a viz nor a best way to tell it. Lastly, once a viz is finished, that isn’t the end. As projects like Makeover Monday show us, once a viz has been made, you can continue to remake it and tell stories with it in different ways.

5. Crank up the tunes

When I really need to focus, I crank up the tunes. Reducing the sound of outside distractions (phone buzzing, colleagues talking, TV blaring) helps me focus on my work. Whatever your favorite kind of music, get some headphones and let it blast! For reference, Dvorak’s New World Symphony (No.9) got me through writing this blog post.

Here's the viz I finished after beating the blank-canvas blues.

1 In this particular instance, I had a fabulously rich data set from the UK Family Food Statistics. I knew this data set had some great stories to tell from articles such as The ODI’s “How British diets have changed since 1974” and The Guardian’s “Goodbye, fish and chips: National Food Survey data reveals changing trends in British dining”. I also looked at this data set at the London Viz Club. But sometimes too many options is as bad as too few; I was paralysed by choice.

Les ressources suivantes peuvent vous intéresser...


where can i get the above data?

Great overview! Sounds simplified but still so difficult to put hands on.

Mohon content Saya dibuka Kimball,,

While some useful ideas are presented, some important points are missing, for example using Tableau as the exploratory sketchpad it's extremely well suited for, and for conducting the basic analyses enumerated in the Data Preparation section.

Does it really make sense to draw things out on paper when Tableau is itself an environment designed for 'drawing' with data? I think in most cases not. Once the basic Tableau skills are mastered and internalized, exploring one's data is as simple as sketching on paper, and it has real benefits over the decoupling of intellectual and cognitive abilities that the chasm between data-disconnected imaginings imposes. Put another way: it's better to actually see the data than to draw pictures of what it -could- look like, which then requires actually looking at it to confirm or reject the drawing's accuracy and/or relevance.

In the second case, there's no simpler way to see data's basic properties than to look at it with Tableau. This isn't data preparation, it's the fundamental operations of data analysis. Want to see the fields in a data set? Connect Tableau to it; although it may help to disable Tableau's auto-naming feature. Want to see what the members of a Dimension are? Double-click it, and Tableau will obligingly list them for you. Want to see a Measure's basic properties? Tableau will happily show you the max, min, avg, sum, etc. with an absolute minimum of fuss. Want to see the distinct values for a Measure? Make it discrete, or a Dimension, and Tableau will list them upon demand.

Tableau was created as the best tool ever brought to market for simple, straightforward data exploration and analysis. Use it this way and your need to draw hypothetical sketches of what things might look like pretty much disappears,and your ability to understand the data expands dramatically.

Ajouter un nouveau commentaire 

Abonnez-vous à notre blog