Video Transcript
Hello everyone. Welcome back to TNT. My name is Hobbs and this a vlog were we talk about best practices here in the data and AI world. Today I am going to be dealing with one of the most popular, almost meme-like things that I see people talking about in relation to data visualization.
Welcome back everybody to TNT. As I said, we are going to be talking about one of the most commonly attacked visuals today but before I get into that:
I have a wide variety of nerd shirts that I really enjoy, and I got this one for Christmas. So, this is the blue screen of death message from Windows 95 for those of you who are interested or wondering what was written on my shirt. Now you can move on with the rest of the video.
So, the thing I wanted to talk about today is pie charts. Pie charts receive a lot of hatred on the internet. But inevitably you are going to be asked to build a report or a dashboard where they are wanting you to build a pie chart. And the best practice is of course:
If you have a choice between almost any other visual in the world and a pie chart, pick any other visual in the world.
But no one really talks about why that is. So why is it precisely that pie charts are not a good data visualization? To understand that you have to step way back in time and imagine yourself as one of your primal ancestors.
Imagine that you are running through the jungle and you’re trying to escape a saber tooth tiger. You’re running along and there’s this log in the middle of the path and you have to clear the log in order to escape the saber tooth tiger. In order to do that, in your brain, you need to see, very clearly, the height of the log in relation to all of the objects around it so that you can jump over the top. And your brain is very good at this.
This is why bar charts work so well. Your brain can almost immediately tell subtle differences in the height of bars in a bar chart. But there is no scenario that you ever would have encountered in life where you needed to tell the difference in the width of a wedge. Which is basically what a pie chart is made out of it. We’re not good at that. Our brain is not optimized for this kind of visual problem solving. So when you have your pie chart in front of you and its got 10 or 12 different colors on it and 3 of them are nearly the same size, its almost impossible to say with certainty which of those is the largest. And this puts you in a scenario where it’s hard for you to communicate the truth about the data because the visual is in the way of that truth.
I don’t want to harp about this forever, there’s plenty of articles online talking about how terrible pie charts are. So, what I would encourage you to do, is the next time that you’re asked to build something that includes a pie chart. Talk about the science of the brain, you can do some research yourself. And then take a pie chart and on screen, in front of people, transform it into a bar chart and show them how much easier the one is to interpret than the other.
To summarize - and then I’ll leave this topic alone and I won’t come back to it I promise - you want to stay away from pie charts whenever you can. Donut charts are slightly better because they are basically a pie chart that you made look more like a bar chart but, honestly just skip that step and move all the way to the bar chart.
Hope you enjoyed the video today. As I said this is a pretty common topic, so I’m not surprised that you’ve heard this before. Regardless, if you disagree, if you’re a lover of pie charts leave me a comment below and defend your position, I’d love to hear it. If you have something you’d like me to talk about, I’m happy to do those things as well.
Additionally, if you’re interested in bringing in someone from [Valorem Reply] as a consultant or to work together with me, that’s the sort of thing we love to do. We love partner with people to help them solve their problems.
Hope you all have a great week!
