Taylor Rodgers
  • Home
  • Books
  • Free Tools
    • Requirement Doc Templates
    • Quality Checking Guide
  • Blogs
    • Data Science
    • R Programming
  • Consulting
    • Statistical Consultant
  • Contact
  • Home
  • Books
  • Free Tools
    • Requirement Doc Templates
    • Quality Checking Guide
  • Blogs
    • Data Science
    • R Programming
  • Consulting
    • Statistical Consultant
  • Contact

Building Data Visualizations When You're Colorblind

12/19/2022

0 Comments

 
Like 1 in 10 men, I am colorblind. I’m a “strong protan,” according to this free online test. I detect too much green light, which prevents me from seeing the full spectrum of reds. Many reds look more like greens to me or less bright than they do to you.

I can still see colors though. I see approximately 150,000 of them, including many reds. That may sound like a lot, but the average person sees 1.5 million colors. So I see an abysmal 10% of the shades that you probably see.

For example, it’s hard for me to see the red coat in this famous movie scene in the picture below. When I glance quickly at the photo, the girl almost blends in with the rest of the people in this screenshot. If I stare long enough at it and zoom in on it, I see a dull red tint at best.

I watched the actual clip while writing this article and, truth be told, I never would've known the girl had a red dress on had a teacher not told me such in high school.
Picture
Another example is the front door to my house, which you can see below. To people with normal color vision, the trim and the door are a different color. Even though I still think it looks beautiful, I can’t tell the difference in colors. Ironically, I painted this myself (my mom had to pick the colors for me).
Picture

Colorblindness Doesn’t Impact My Day-to-Day Life Much

At worst, colorblindness is a minor inconvenience in my day-to-day life. I don’t think it’s a real disability, because its impact is marginal.

I also think it’s disingenuous to solicit pity or concern for being colorblind. I do not want messages or comments about how sad it is and I dislike those “watch a colorblind person wear corrective glasses and see their daughter’s green eyes for the first time” viral videos. I always thought those videos were patronizing and condescending.

Being colorblind does impact my life in some minor ways though, such as the clothes I wear. My favorite shirts, hoodies, jackets, etc. are usually a shade of blue. Turquoise is my favorite color and I would paint my house that color if I could.

It also kept me from ever seriously considering joining the military when I was younger, as most military jobs I found interesting required normal color vision.

I would say the biggest inconvenience I ever faced in my personal life related to video games – mostly Red Dead Redemption. It’s one of my favorite games, but whenever I played it, I had trouble using the navigation menu in the bottom corner because the red line blended in with the map. It didn’t stop me from enjoying the game though. Later, Rockstar Games actually offered a “colorblind friendly” mode in the sequel Red Dead Redemption 2 for defective eyes such as mine.

Colorblindness Does Impact My Data Science Career

Colorblindness did pose challenges in my career. Less so now, but when I worked as a data viz developer, I had to work around being unable to see colors correctly.

Color is an important part of data visualizations. Many people who don’t do that for a living put little thought into it. But a subtle color strategy really does help people see the key insight in a visualization.

Colorblindness obviously makes this part more challenging for me. 

Before deploying an application or giving a presentation, I’ll have to tap on a co-worker’s shoulder and ask them to verify that it looks okay. Thankfully, most coworkers have never complained about this. Something I do appreciate. :)

One boss discovered that I was colorblind. He sighed, shook his head, and joked, “Of course, I would be the one guy that hires a coloblind report developer." If you think that’s an insensitive joke, please don’t. I think it’s funny too.

A common problem I had was with Tableau’s default color palette. Down below is a screenshot from Tableau’s interface. If you use the colors below in your data visualization, just know that nearly every color looks like another color in the same palette. Three colors in particular look identical to me.
Picture

Colorblind Friendly Palettes Don’t Always Look Good to Me

Many data viz tools offer a “colorblind friendly” option. These have improved in recent years, but I should tell ya, many look awful to me.

You might say, “Well, you’re colorblind! How do you know if we find them ugly?! We find them beautiful!”

But if your goal is to be more accommodating to colorblind people, how does replacing beautiful, but similar looking colors, with these barbaric, disjointed palettes accommodate us?

Instead, here’s what I do to use color effectively.

Tip#1: Use Brand Colors

Whenever I build an application for a company, I’ll go and find their brand guidelines. Usually a graphic designer more talented than myself will have created a whole color palette. I’ll use the colors from that palette that make sense for the data visualizations.

That eliminates the concern for me that I’ll select a “bad color.” If it’s bad, it’s the graphic designer's fault. :)

Tip #2: Use Labels Instead of Color

Generally, I don’t have to worry about colors on a plot clashing, because I use one color at a time for a data viz. The way I distinguish categories is through labels.

That's what I did on the data viz below from a portfolio app I built years ago. There's no real need to distinguish color with these categories. It's better to label the categories instead.
Picture

Tip #3: Use 2-3 Colors Very Different From Each Other

Sometimes, it makes sense to use two colors. If I do, I make sure to pick ones I can distinguish. So if I pick a blue, I don’t make the other one a purple, because purple (having red in it) makes it less distinguishable from a blue in my eyes. I might use vivid red or black instead.
Picture
In general though, once you go beyond three colors, you have too many.

Best Tip Overall: Use a Single Color + Grey Scale

Lastly, I tend to use less color overall. I’ve become a fan of grey scale or very subdued colors. I find it’s easier to read through a data viz when it has less color, cause I pay closer attention to the labeling and the size of the shapes. This might be my colorblindness, or it might be because I’m a fan of apps that let me read the info, as opposed to visualize it.

That doesn't mean that my apps have no color in them. Rather, I use color to highlight what's important for the user to see.

For example, I use orange (and a dashed line) to distinguish one different type of data point from the black line down below. Orange is the only non-black and non-grey color on this entire visualization, which helps draw attention to it.
Picture
And on the next charts, I use blue to highlight what the user selected in the drop down filter. That helps them see what  is important to them, while the rest of the bars / lines are black or grey.
Picture

Final Thoughts – Colorblind Friendly Is Really Just Good Design

As you may have noticed, my way of designing good data visualizations while colorblind sounds like just good design tips in general.  That's cause they are!

You don’t want to overwhelm the user with unimportant info. Too many colors is too much information. Using fewer colors and using more labels allows you to circumvent the problems a colorblind person would experience, without even worrying about whether it's "colorblind friendly."
0 Comments



Leave a Reply.

    ABOUT

    A blog about the non-technical side of data science.

      SUBSCRIBE

    Confirm

    ARCHIVES

    April 2022
    March 2022
    October 2021
    September 2021
    August 2021
    March 2021
    February 2021
    January 2021
    November 2020
    August 2020
    June 2020
    May 2020
    April 2020
    February 2020
    January 2020
    December 2019
    November 2019
    October 2019
    September 2019
    August 2019
    July 2019
    June 2019
    May 2019
    March 2019

    RSS Feed