Hi! I'm one of the shortlisted in IIB, with DrWhy.AI Data Visualizations project.
It is a collection of tools for Explainable AI, created by Przemyslaw Biecek and MI^2 Data Lab. I’ve been asked to redesign the DrWhy visualizations to be more understandable, useful and engaging for the users.
These are my personal favorites: [0]
Such Genomic data representations are so rich, you can see patterns and tells if a tumor (when applied to cancer) has an unstable genome or a P53 mutation, etc. You would almost run CNN based image analysis on them, if that wasn't so silly (It would be more similar to how us humans view the plots though).
I have to admit I was pleasantly surprised by these. A lot of these "information is beautiful" things used to put the emphasis more on beauty, even at the expense of not being informative.
These seem to be a good balance of both information and beauty.
Out of curiosity, I looked at the older years, like 2013. Maybe it's just my confirmation bias (I believe/hope that, as online dataviz matures, it trends towards simplicity and scalable patterns) but the older shortlisted entries seemed more bespoke (in a convoluted way) and more infographic-ish, i.e. less useful than the 2019 shortlist entries:
My favourite one shows the orientations of all airport runways [1]. Since they runways align with prevailing wind directions, it's effectively a map of the winds.
It is a collection of tools for Explainable AI, created by Przemyslaw Biecek and MI^2 Data Lab. I’ve been asked to redesign the DrWhy visualizations to be more understandable, useful and engaging for the users.
If you want to use it, here's the code: https://github.com/ModelOriented/DrWhy
If any questions, feel invited!