For week 17 we teamed up with Tableau to look at LinkedIn’s ranking of the skills that are most in demand. From a selfish perspective for Makeover Monday, it’s pretty amazing to have such close support from Tableau. What’s even better is that Tableau realises this isn’t a “Tableau” project; it’s a data visualisation project. And they’re ok with that. Just like us, Tableau wants to see people using data to communicate effectively.
WORKING WITH RANKS
The data provide was nothing more than a ranking of the jobs. This can be quite tricky for making comparisons because even though one job might be ranked 2nd and another 5th, we really don’t know how big that gap is with seeing the raw data from the survey. It would have been great if LinkedIn provided the number of responses to each question. Then you’d be able to measure the true gap between responses. As you’ll see below, though, you can still come up with some great visualisations with such bland data.
CONTEXT IS KEY
I beat the context drum constantly. Context helps provide your audience with a simple way to make comparisons without having to do any mental gymnastics. There are many ways to add context: comparison to prior year, comparison to an average, comparison to a target, etc. Then visualise that through methods like highlighting or indicators or reference lines.
In this example, MS used highlighting.
Meanwhile, Georgia Chen used arrow indicators.
Both are very effective at focusing the audience’s attention on what’s important. Well done ladies!
WORKING WITH INCOMPLETE DATA
This data set was kinda crappy in that there were tons of holes in the data. For example, in 2016 there are far fewer jobs listed than 2014 and 2015. I’d warn people against using incomplete data for comparison purposes. I saw way too many bump charts that would have a dot by itself or connected for only two of the years.
There’s no real good answer to this, however, I’d like to offer up a few suggestions:
- Only include roles that existed in all years
- Exclude 2016 since it clearly didn’t include all of the roles in the data
- If a job was in the top 10 for one or more years, but was outside the top 10 for another year and you’re limiting your viz to the top 10, make sure you don’t make it look like that job didn’t exist in the other year(s). This can be misleading.
- Consider not “connecting” the years to avoid #3.
USE COLOR TO ADD MEANING
Next time you add color, think about what value it’s adding. Does it improve understanding to color every job differently? I doubt it. What if you remove color? Does it take away from interpretation? What if you use color to highlight what’s important? Color should be used to add context to your visualisations. If it doesn’t add context, consider removing it.
Great work this week with a tough data set. With that, here are some of my favorites from week 17.
Author: Matt Chambers
Link: Tableau Public
What I like:
- Great example of a ranked dot plot
- Fantastic hover actions
- Nice use of only two colors to indicate whether a role is above or below the global average
- Row banding helps your eye follow across the visualisation
- Including a reference line for the global average for each role
- Sorting the global average to the top then the rest of the countries alphabetically
- Including instructions
Author: Neil Richards
Link: Tableau Public
What I like:
- Great to see Neil iterate on this and make some simple changes that made it look better
- Nice use of highlighting to add context across the countries
- Good instructions that explain the viz
- Minimal use of color
- Using filled circles and lines to connect makes it look very sharp
- Design makes it easy to see the years when there are gaps in the data
- Alphabetical sorting of the countries from upper left to lower right
Author: Korey Klein
Link: Tableau Public
What I like:
- Slope graph shows the changes very well; excellent chart choice!
- Using color intentionally and only where necessary
- Nice labeling
- Very easy to understand
- Great title
- Focusing on a single job role tells a nice story
- Overall simple, elegant design
Author: Simona Loffredo
Link: Tableau Public
What I like:
- Sometimes a simple table is all you need; this is one of those examples
- Including the country filter makes it easy for the user to pick what’s important to them
- Lack of color
- Super simple and easy to understand
- Making a single year the focus of the story; if you’re looking for a job, recency is most important.