Redesign of a truly bananas chart

On Twitter, I stumbled upon this horrendous 3D bar chart. When looking at the data, it might have been made in 2005. Data visualisation as a skill was yet to be defined. Boy, we have come a long way, but this was even at the time truly bananas.

Normally I describe in detail about what improvements can be made, this time I trust my audience to recognise the many pitfalls. I will however explain my redesign choices after the visual.

Watch out though, PhillipDRiggs warns you in his tweet for “post-traumatic viz syndrome”.

Redesign choices

To me, this had to be a line chart because we deal with years in time. Also, if you want to plot everything in one graph, lines can be convenient because they take up little space.

Let’s see if that works.

First iteration

That’s better already, but still your eyes have to switch a lot from the legend to the plot and back again. One thing you could do is plotting the country labels inside the plot and remove the legend, but there’s a lack of space.

When trying to squeeze three variables, and in this case ten countries, in one plot, things get messy. Welcome to small multiples, sorted from highest to the lowest mean.

That’s way better already. You can easily compare countries because they have their own little plot and they’re sorted, giving the reader instant gratification or information if you will. Per country you can scan the trend through time as well.

For the final version, I think that an area chart works better because it represents the magnitude of tonnes of bananas better. And maybe even visualise the trends more precise because the big contrast of the background it creates these sharp edges.

Small multiples. Left: line charts. Right: area charts and final version.

This is part 9 of the series Data Visualisation Redesigned for the Better.

You can find the code behind the redesign on the Colourful Facts Github repo. For more redesigns and other data journalism/visualisation related articles, go to my home page:

Do you stumble upon a crappy graph? Please let me know! Cheers 🙂