How to Create Good Visuals

[This article was first published on R – Vertabelo Academy Blog, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

In this article, we’ll take a look at guidelines you should follow to create compelling visuals. Our goal is to learn how to effectively convey information through graphics.

Have you ever looked at raw data—spreadsheets of stray numbers—and struggled to make sense of it? We’ve all been there, but it’s no surprise—because the human brain processes visualizations and images 10,000 times faster than raw data. In fact, 80% of the information we absorb comes from visuals, and the remaining 20% is text. As such, it’s much easier to understand data when it’s represented visually.

Regardless of what field you work in, you’ll need to know how to effectively and accurately convey information to your target audience. This is the key to success in today’s data-driven business world. But as with most things, there are both good and bad ways to go about creating visuals. Consider the chart shown below.

Clearly, it’s a pie chart, but what conclusions can you draw from it? Not much, really. The legend isn’t very informative, and there’s no title. Hopefully, no one will present this kind of chart in real life, and we realize it’s a bit of an extreme example—but you get the point. We won’t delve too deep into the specifics of the different types of charts and graphs you can construct, but if you’re interested in learning about those, you should look into Vertabelo Academy’s Data Visualization 101 course.

Okay, let’s get back to the pie chart above. Below is another visual that uses the same data:

I think we can agree this is certainly a step up from the first one we looked at. For one, the title tells us what the graph represents: the number of new taxi drivers each day, over a two-week period. The lines allow us to trace the increasing/decreasing trends. Moreover, the axis labels tell us what information we’re looking at: the x-axis represents the dates, in increments of one day, for the year of 2015. The y-axis represents the number of new taxi drivers, ranging from 0 to 70.

Can you believe those two graphs use the same data? It’s true! Clearly, quality is an important consideration when constructing visuals—how useful is a graph if nobody can make sense of it?

Below are some guidelines you should follow to create good visuals and convey information effectively.

Good data preparation

The first important step is preparing an accurate data set that you can use to create your visuals. Quite often, this is actually the hardest step because it involves data extraction, cleaning, loading, merging, imputation, and a lot more.

Once you’ve prepared your data, you need to decide what type of data you’re working with and how you plan to present it—do you have discrete, continuous, or categorical variables? And which chart is most appropriate? Note that Vertabelo Academy also covers data variables types in its Data Visualization 101 course.

Know your audience

Consider your audience—who’s going to be looking at your visuals, and for what purpose? Visuals intended for a company management board will certainly differ from those intended for business analysts or IT experts. Analysts look for detailed stats backed up by raw numbers, whereas IT personnel require technical specifics. In contrast, a management board will want to see facts, impacts, and main conclusions with fewer details—because their primary goal is decision making.

It’s vital that you know your audience and adjust your visuals accordingly. This not only makes it easier for your audience to understand the information you’re presenting, but it also shows that you understand their needs.

Emphasize important information

Emphasize what’s important! It sounds fairly obvious, but presentations are too often cluttered with irrelevant details that make it hard to see the bigger picture. Consider what you’re trying to convey, and emphasize the most important information.

For example, if you’ve created a bar chart of earnings per year and want to highlight a particular year as the most profitable, consider brightening its color to distinguish it from the others. This immediately makes that element the center of attention, and it becomes that much more memorable. Take a look:


Don’t forget about accuracy

Visuals can be easily altered or presented dishonestly to suggest something is true when it isn’t. Take a look at this example of a misleading graph:


The problem is subtle. At first glance, it looks like the earnings were indeed much higher in 2016 than in 2015. But if you take a closer look, you’ll notice that the y-axis starts from 340 and not from 0. The smaller scale here exaggerates the difference between earnings for 2015 and 2016. In reality, that difference is only very slight. Here’s the real representation of this graph:


Notice that these two bars are about the same height—it’s the same data, but the way you present it can lead to a different interpretation. Always try to present information accurately and honestly.

Make your graphs readable

Readability is one of the most important considerations when designing a visual. If people can’t bear to look at your graph or simply don’t understand what it represents, you won’t be able to get your message across. Let’s look at some techniques that you can use to improve graph readability.

Colors

Recently, one of my clients (for whom I’ve worked as an exploratory analyst for some time now) wanted me to prepare visuals in different colors. I didn’t want to argue with him, so I did as I was told. But here’s the thing: more colors don’t always make your graphs better. It’s generally recommended that you stick to 3–4 colors. Any more, and your audience will be overwhelmed and confused. Take a look:

It’s almost blinding, isn’t it? The chart above depicts the world’s most populated cities. Each city is represented with one bar, and each bar is colored differently. It’s pretty hard to look at this for more than a few seconds because of all the bright colors. Try to not do that—don’t represent your bars with different colors if they are already labeled. The colors don’t offer any additional insight; the same information can be conveyed much more clearly with just a single color:

Be conservative when it comes to colors. Also, avoid combining red and green colors. Why? Because color-blind people can’t really distinguish those two colors very well in combination; it’s good to choose another combination of colors or add some labels when preparing graphs for such an audience. If you aren’t sure what colors you should use, take a look at the how to use color exercise in our Data Visualization 101 course.

Choose the right visuals for comparing values

Often, visualizations are used to compare values among groups or categories (earnings per year, population among cities, salaries by company, etc). You need to choose the right visuals for your purpose. The best charts for comparing values are bar charts and line graphs. For example, city populations can be easily visualized with a bar chart, such as the one shown below. And if you additionally also sort the bars in descending or ascending order, it makes it easier to spot certain patterns.

Ordering really contributes to readability. Notice how we sorted our bars in descending order to present the most populated cities first (thereby emphasizing them).

3D charts are often useless

Adding a third dimension to a chart is, in most cases, unnecessary and useless. If 3D visuals don’t add any valuable information to the visual or contribute to readability, don’t use them. Fancy charts are not necessarily good charts.

Rearranged icons and labels can help convey your message

Small things can make a big difference. Modifying labels and adding them to different parts of graphs can improve readability. For example, switching to horizontal bars and adding a population number into each bar, increasing the font size of the title and axis labels, and bolding the title are all good improvements:

Just like colors, titles are important to consider in data visualization. The “Add a chart title” section of our Data Visualization 101 course teaches you to create good titles using the R programming language and the ggplot2 package.

Less ink, clearer image

Our previous visual was mostly blue. Here’s another important rule to remember: if your colors comprise more than 60% of the visual, change the plot type—use something else! For example, replacing the bars with points or lines can make a big difference—there will be less color on the screen, but you will still get the same quality of information. Below are the most crowded cities represented with points instead of bars. What do you think—which visualization looks better?

Less is more — avoid complexity

Last but not least—don’t go overboard with your visuals. Additional illustrations, arrows, unnecessary text, and dark grid lines and background colors (“a lot of ink”) are the so-called chartjunk of charts and graphs—they contribute nothing to the plot and are often distracting and confusing. One of my personal favorites is the “243% baby boomer” illustration:

Here’s a question: Why is there a picture of a human body on the screen? Does that help you understand the information any better? Of course not. And what do the percentages represent? Because when you sum them up, you get 243%—and we all know it should at most be 100%. Those percentages represents the results of a survey, and they should’ve actually been visualized in separate graphs. Everything presented on one screen is unclear and misleading.

Another graph also caught my eye:

Too much, don’t you think? This is a great example where the information could’ve been conveyed more clearly through a simple bar chart, without all the distracting visuals. Remember—simplicity improves readability.

Conclusion

Creating visuals is not an easy job. In this article, we covered some of the best practices you should follow when creating good visuals. To briefly summarize, consider the following questions when designing a plot:

  • What does my data tell me?
  • Who is my target audience?
  • What message(s) do I want to convey?
  • Is there a certain piece of information that I want to emphasize?
  • Is my visual readable? Can I understand it by simply glancing over it?
  • What is the simplest way to visualize my idea?

If you know the answers to these questions, then you’re good to go! This article should help you create legible charts with strong and clear messages.

Now roll up your sleeves, ‘cause it’s time to get to work! Our Data Visualization 101 course guides you in using R and ggplot2 to create beautiful visuals. You’ll even get a certificate of completion when you’re done so you can show off your knowledge to the world. See you there!

The post How to Create Good Visuals appeared first on Vertabelo Academy Blog.

To leave a comment for the author, please follow the link and comment on their blog: R – Vertabelo Academy Blog.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)