Join Bill Shander for an in-depth discussion in this video Color, part of Data Visualization for Data Analysts.
- Color is incredibly powerful and equally as dangerous. It adds visual interest to your visualizations while potentially increasing readability of your data, but just as it can make it more readable and beautiful, color can just as easily make your visualization a hot mess. Following just a few simple rules can keep you from making a visual disaster. There's a lot of established thinking about how to use color in design. You could spend plenty of time learning about how to combine colors and what colors mean and human psychology and cultural references to color.
In fact, you can do some of that here on Lynda.com. Anyway, the main thing you need to understand is that color is probably the hardest thing to learn. You might be able to learn it, but it's definitely not guaranteed, and, unfortunately, many of the tools you use, yeah, I'm looking at you Excel, they don't help, because they use default color schemes that can lead you astray, even before you blink. So, the first place to start, if you can, is to hire a great designer with a fantastic color sense to establish a color palette for you.
If you work at a large corporation, odds are you have a color palette as part of your brand. Find the brand guidelines. Spend an hour talking to your branding department and learn how to use the colors. This is time well spent. Let's assume that maybe you can't afford to hire a world class designer, and maybe you work for a company that doesn't have a color palette, so you have to pick your own colors. Okay, don't freak out. It's okay. If you must, you can use tools to keep you honest. So one of the tools that I use is this website, used to be called Adobe Kuler, and it's now called Adobe Color CC.
Essentially it's a tool that will general palettes for you. And so, I'm just going to scroll down a little bit so you can see the whole thing here. Here we are looking at a five-color palette. It is following certain rules. So this one is an Analogous palette, meaning the colors are all in an analogous family, right? They're sort of close to each other on a color wheel. There are different types. I can generate Monochromatic palettes, Triad palettes, and sort of different words for things that you may not know what they mean, but you can see what they do as you click into the different choices here.
You can also roll your own palette. You can actually grab one of these things and move it around, and you'll see that not only the color you're moving changes, but the other colors change. So it's sort of following some color theory behind the scenes to make sure it's picking a five-color palette that works together. So it's a really interesting tool, one that I like to come and play with a lot of the time. You can also save your themes here, which is kind of nice. Another website, another tool that I really like for generating palettes is this one, i want hue, and as you can see over here, this generates colors for data scientists, okay? It will generate and refine palettes of optimally distinct colors.
So what does that mean? It means that if I ask for a seven-color palette and I hit Reroll palette, it's going to generate a palette of seven colors that are optimally distinct, meaning that they're easy to distinguish from each other. They're all easy to distinguish from all of the others, at least as much as they possibly can. I can generate a 12-color palette this way. Now, I wouldn't recommend this. You want to minimize your color use, but it's a really neat tool to play with to pick palettes. One thing you really have to keep in mind, color blindness. So, if you look at this image, all of these circles have numbers in the middle of them, and the vast majority of you listening to me right now can see those numbers, but 1 to 5% of you probably can't.
The most common form of color blindness is called red-green color blindness. So people with this disability struggle to distinguish between red and green. So when you design a chart using red and green to indicate bad and good, be aware that those cues you're assuming will be instinctively understood may not be by a significant chunk of your audience. Luckily, there are tools for that as well. This is one of the most commonly used, COLORBREWER, which is a website that will let you pick different color schemes that are pretested for color blind friendliness.
And again, I have different types. I have sequential hues, meaning they all sort of go in order and build upon each other, or diverging palettes, divergent colors, and these are sort of designed with maps in mind, but they work really well for charts as well. So this COLORBREWER website is a great website. There's also this tool that I really love. I have this installed on my machine, called Color Oracle. It's an application, and it will let you test your own designs. So it lives up in the task bar, and all you have to do is sort of open up Photoshop or Excel, whatever you're working in, and go to ColorBrewer, and it will show you what your design, what your chart looks like as though you had a form of color blindness, and it has multiple forms of color blindness you can test.
Another issue is, like in fashion, there is such a thing as too many colors. As I mentioned earlier, that 12-color palette. It's really hard for people to see that many colors and parse them. So even if you're using a tool with good default colors, the next challenge is that there's a limit to how many you can distinguish before they start to blend together. So if you're using more than five to seven colors, you really do risk confusing your audience. I've never seen a 10-color palette that was anything but overwhelming. Now, if you can do what you need to do with one to three colors, all the better.
Less is more when it comes to color. So yeah, this chart's a great example. Just because Excel, for instance, tries to assign a color to every single thing, just because you can use many colors, doesn't mean that you should. Only use color to express something, to make variations stand out, to categorize things that need categorizing. Use as simple and minimal a palette as possible. And so, for instance, here's that same chart. This is the exact same chart as the previous one. I've eliminated all of the colors, and I'm only showing the things that matter.
So, for instance, the year 2014 in this case is probably interesting. There's something going on there. That's the current year when this chart was created, and then there's an outlook for 2015 being illustrated. Color can make those things stand out, sure, but we didn't need as many colors as were in that previous version. Color is scary. Every time I have to start a new project, it makes me as nervous as anything else, because it's so hard to get perfect, but if you follow these rules you increase your odds by a huge percentage of creating something that looks great and that communicates what you want to communicate.
- Why visual communications matter, and how they work
- Communicating via story
- Communicating with color
- Using legends and sources
- Sketching and wireframing
- Rethinking slides, charts, and diagrams
- Rethinking your templates and brand guidelines