Quote:
Originally Posted by Bingo
Team Chart:
Flames 5on5 / 60 minutes VS the 16th best team (bubble) in each category.
Really points to where the team needs to improve, and the fact that they've taken steps in the right direction.
[chart]
|
I'm totally on your side as far as the importance of data and analytics in hockey, but I've gotta be honest: that's a bad chart. If you're trying to communicate something through data visualization it's your job to make it easy and intuitive to interpret.
Specifically:
- Lines shouldn't be used to connect discrete, independent data points; they should be used to connect samples of a continuous value. The use of a line implies that you can interpolate between values on the line. So if a line chart measuring temperature has a data point at [2PM, 12 degrees] and at [6PM, 6 degrees], you can interpolate and say at 4PM it was probably ~9 degrees. In this case there's no intermediate value between [Game 2, 8 Scoring Chances/60 above median] and [Game 3, -8 Scoring Changes / 60 above median]
- Different measurements sharing the same axis. The bars being right next to one another implies they should be compared, but they're measuring different things. Visually, seeing the orange bar above the silver bar feels like it should mean something, but it's actually meaningless because Corsi events and scoring chances are different things. If you want to compare them on the same chart, some sort of normalization should be applied first (Maybe plotting the z-score of that measurement).
In general, I've found it much more effective to use
Small Multiples when you've got a bunch of related, but different data to display.