Master your data visualization skills (4 tips)

Jan 14, 2019·7 min read

“Just find the viz you like and copy it” — have you seen such advice? I have. After one of these tips I saved a lot of dashboards and google pictures with nice colour combinations. At that time, I did not realized that colours and shapes (visual attractiveness) are not the only things you need to know in order to become a visualization master. You should also attract your audience’s attention, to know exactly how and where to place accents and other subtleties. In this article I will cover main things you need to know on this subject and I hope you will find something new for yourself in order to make our world clearer and more beautiful!

Let’s draw 2 axes, where Y — visual attractiveness and X — functionality. Take a look at the diagram below and 4 visualizations of the same dataset.

Looking at each viz can you (and how quickly) answer the following questions:

Where are the most expensive taxis?

2. Where are they the cheapest?

3. Where are taxi costs about the same?

4. Is there France somewhere in the chart?

4 tips to master your visualization skills:

1. Make analysis, ask yourself:

Who is your audience and what do they want to know?
What are the main ideas you want to show them, what do they have to release or do after viewing your visualization?
What additional data do you need to show in order to prove your point?

All your answers are the basis for visualization, and at the same time this is a good way to check your final product.

Do not hide facts that do not confirm your idea, but also do not show the audience all your exploratory analysis and a long way to the main point (only if you need to show them you are truly professional). Keep only the main points (explanatory analysis) and proceed to the next step.

2. Choose the best way to visualize data:

“When it comes to the type of graph, you should always use whatever will be easiest for your audience to read.” (Cole Nussbaumer Knaflic, “Storytelling with Data: A Data Visualization Guide for Business Professionals”, p. 120)

There are types of visualization that require more time to perceive, because we read them (verbal perception): text, tables. Other graphics we perceive faster (visual perception) — those where there are colours, shapes, sizes. This does not mean that you should stop using tables, just keep this in mind when you make a choice. Let’s look at the most popular graphics and where to use them:

There are many other ways to visualize your data, just take two minutes to analyse them before using. Do they reflect the main idea in the best way? Can you imagine a simpler and and more intuitive way to show your data?

It is better to avoid/use less often some types of graphs, because they are meaningless or have many limitations: pie chart and donut (read more here), scatterplot — it is usually used in science to show the dependence of two measures. 3D — it’s hard to imagine when this kind of graphs can make real sense. Rainbow graph — when there are too many categories try to group them or create several diagrams.

3. Cleaning

“I would have written a shorter letter, but I did not have the time.” (Blaise Pascal, 1656–1657)

Everything that you have on a dashboard takes your audience’s attention, but our attention is VERY limited. Help your audience to easily understand the data — remove everything that defaults and/or does not have much importance. Start with a white sheet and continue cleaning your viz all the time as you create it.

What could be default or unimportant? For example, underlines, frames, visible lines, where they are not necessarily needed. Alignment is an invisible border, but still it is a border, so there is no need to add unnecessary lines or marks to the chart. Another example: value labels with lots of details, both % and absolute as well as the name of a group and subgroup. Move some of the values into comments, tooltips, or a separate table if you really need them.

4. Emphasis (design tips)

You have already identified the main ideas of your visualization, now you can determine what your client will look at first. You can also determine what he will notice afterwards and so on.

There is a rule in the design: if something is not quite the same, make it completely different. For example, comparing profits and costs you should not make one line blue, and the other dark blue — make them in different colours.

Below are ways to highlight and logically emphasize elements in your visualization:

Colour: the default colour combinations in Tableau are fine, but sometimes we need to find the right colour combination on our own. There is an easy way to find the right colours with a colour calculator based on simple rules of colour harmony. Use colour shades if the current colour combination resembles Christmas or a toy store.

It is important to know that warm tones are “stronger” than cold ones (they take more attention). So, for example, comparing year to year you should not make the current year blue and the previous orange — all attention will be given to the previous year. Do not be afraid to use grey colour, it helps to place accents!

Repetition: repeated design elements (colour, font, size, alignment, emphasis) in your visualization give it integrity and functionality.

For example, if you repeatedly show profit, make it the same colour in all charts, so your audience will intuitively understand what they are looking at. Another example is subtitles, it’s better to use one font and size for them, so the structure will be better.

Font and Size: there are rules on how to combine fonts, but it is much easier to use one font with variations of bold, medium, regular, light. When you choose the text size remember that difference of font size in 2pt is not a difference for our eyes, so make it REALLY different.

Underlining (except hyperlinks) usually complicate the reading of the text, so it is better not to use it. There is also a rule of punctuation: the punctuation has to be in the same style (font, colour, size) as the text before it. Example: text, not Example: text.

Proximity: we perceive close objects as related, and remote objects as unrelated. Therefore, for example, you should not put group names too far from the histogram, because we do not see that they are related:

Alignment: everything should be aligned and sorted. The easiest and most reliable way is to use left alignment. It is also possible to use other alignments, but you should know how to use and combine them in a right way. The use of diagonal alignment is generally meaningless.

By the way, all elements on your dashboard should be aligned with each other, nothing should fly randomly or stand out from the general structure, so the work looks neat and professional!

Placement: designers know where to place an advertisement or the most important information, it is also useful for us to know. According to eye tracking studies (tips specifically for Tableau here), we read the page in a very similar way:
we basically start from the upper left corner
we always first we see what is on the top
we spend most of the time on the left side of the page

https://vanseodesign.com/web-design/3-design-layouts/

Remember this and use to create your perfect viz.

What if your audience prefers Excel spreadsheets? Just give them a table, and also add a perfect visualization. One day they will realize that good visualization is priceless!

Most of the ideas described in this article (and many other examples) can be found in 2 brilliant books:

“Storytelling with Data: A Data Visualization Guide for Business Professionals” by Cole Nussbaumer Knaflic
“The Non-Designer’s Design Book” by Robin Williams.