The Art Gallery Guardian

Misleading while being honest

Given a set of data AA, we can plot it on a radar chart. One can permute the axis to make sure the area of the radar chart is maximized. This was explored in a previous article.

More interesting problem. Given two sets of data AA and BB, we are interested in finding a common radar chart, that make AA look as good as possible compared to BB. We might want to optimize the area ratio, area difference, or something else. Again, we are thinking of permuting the axis of the radar chart.

I once mentioned this problem to R Ravi, and he suggest I could ask the same question for all kind of different graphs. How to mislead people with graphs while being completely honest? Indeed, this looks like a fun research project. There is a wikipedia article completely devoted to it. In my previous post, I've discussed how to fitting two seemingly not that related data points through simple transformation.

I'm interested in are algorithmic problems where one want to compute the most misleading chart, I think it would be a great fun project.

Posted by Chao Xu on 2019-04-06.
Tags: data visualization.