Picking great colors for your visualizations can make the difference between a ho-hum or confusing visualization and one that really makes the point.
Today I've been experimenting with Calendar Heatmaps in Tessitura Analytics. Below are two color palettes on exactly the same data. The only difference is the color palette I've chosen to communicate the values.
The Colors palette on the two charts makes a big difference in the clarity of the communication.
On the left, I chose a pallet like this
This seemed to be a reasonable set of color in the Color picker. The colors for the highest values were warm red-orange colors at the top to Lowest at Bottom with dark blue colors.
However, the resulting heat map makes it hard to understand that most weekends we establish more new memberships more consistently. When I just change the colors to the palette to the below. I end up with the heatmap on the right. Same breakpoints. Just different colors. Here are the colors I've used.
Now we see a weekend pattern and an interesting set of weekdays where LSC get more new memberships. (Hint these busy days include, MLK Day, Presidents Day, and Spring Break Week.) These show up very clearly.
So picking good colors can be hard. There are lots of issues to consider. Here is a link to an article about picking good colors for your data visualizations
https://blog.graphiq.com/finding-the-right-color-palettes-for-data-visualizations-fcd4e707a283
There are lots of others out there.
For the calendar heat map on the right above, I ended up using ColorBrew2 to select my colors.
I then pasted the colors hex values into the color picker in Tessitura Analytics
And presto, a considerably more communicative calendar heatmap.
I would love to see other folks chart makeovers. What did it look like before you picked a great color palette? What does it look like now? How have the colors helped communicate the message more clearly?
--Tom
Though I don't have any specific examples to share here, I would be remiss if I didn't steer everyone toward the work of Edward Tufte, the master of the principles of elegant data viz. Not sure if he's come up in your conversations before. He has some very strong opinions on this subject -- which you may or may not agree with -- that are nonetheless very thought-provoking.
If you're not familiar with his work, I highly suggest checking out his wonderful books, or signing up for his one-day course if you can. https://www.edwardtufte.com
I see that he is holding classes here in Brooklyn, NY later this fall.
Buongiorno amici,
Ive had trouble specially with scatter maps often sitting in front of the chart for 20 minutes at a time before i go back to my red to yellow (min/max) scheme. The big thing for me is that red is the only colour i find with enough colour contrast to make low density regions pop. And there is a psychological predisposition to instinctively seeing warm colours as max.
Ciao
H
I have been familiar with his books for quite some time, and recently finally got to take the course in Princeton in early August (the Brooklyn dates weren't convenient with my summer travel schedule!).
I'd recommend starting with The Visual Display of Quantitative Information, a classic IMHO. My personal fave is Envisioning Information, but for the purposes of this particular discussion, the best content might be in Visual Explanations: Images and Quantities, Evidence and Narrative or Beautiful Evidence.
I also felt like the most powerful principles Tufte discussed in the class came from the documentary he produced about Inge Druckery, which you can watch here: http://teachingtosee.org/film/TeachingToSee.html That masterful discussion of maps of the Swiss Alps at the 31 minute mark really encapsulates what can be done with color, labels, and visual data in an elegant way.
Can I just say that this is wonderful. I was just about to do a Data Visualization topic for our fortnightly lunchtime lectures and this has me really wanting to dive in head first. Not just starburst and onion graphs now!
Heath Wilder,
If you can make analytic Coffee this Friday (Saturday Morning) maybe we can do a bit of show and tell.
That should be OK. I start late this Saturday. Do you want to talk Data Visualization?
Let's put it in as a topic and see what the gathered hive mind thinks.
Do you want to talk about it as visualization makeovers?
Or color in visualizations?
Or _____________?
Interestingly this article was just floated our in the PASS insights newsletter
https://www.tableau.com/about/blog/2019/9/3-storytelling-color-tips-improve-your-data-visualization
Heath Wilder
can you share the link to the PASS insights newsletter you got this from?