I love this example because it's a good illustration of the fact that one cannot simply learn the lesson from Anscombe's quadrants and think that increasing granularity always leads to greater transparency/faithfulness. Too many details can hide a signal as much as too few!
I wrote a post with my presentation/paper highlights from the fabulous IEEE VIS 2025 conference I attended two weeks ago in Vienna. This list is based exclusively on presentations I was able to attend. There are many more great papers! Let me know if there's anything you really like.
I’ve compiled a list of papers in my newsletter to get up to speed with the latest research on generative AI for data visualization. I hope you'll find it useful! filwd.substack.com/p/a-reading-...
Maybe good for prototyping?
I had a great chat with Richard Brath on the use of AI tools to visualize text data. Have you ever seen his experiments? Really fascinating! You can watch our chat on my Substack: filwd.substack.com/p/visualizin.... Make sure to check his fabulous blog posts too! richardbrath.wordpress.com.
The slides for my free webinar on "Misleading Data Visualization" are ready! Join me this Monday, May 12, at 12 p.m. ET, to learn my framework for detecting misleading visualizations. (Find directions on how to enroll in the thread.)
Oh, sorry ... I did not realize it!
Question: What process do you follow to add icons to your data visualizations, diagrams, tables, or infographics to convey a concept? I think icons can be very powerful to quickly orient people to the right information, but also incredibly hard to get right. What strategies and tools do you use?
Yes! You can try! It takes probably less than 5 minutes! :)
New experiment to see if ChatGPT can spot data interpretation errors in charts. I tried with four classic misleaders. It got two out of four right. See the details in my newsletter post.
What do you think? What else should I try?
We are organizing a Vis+AI meetup at Northeastern University on April 3rd. Join us if you are in the area! P.s. We are trying to see if we can also stream the event online. www.eventbrite.com/e/visai-meet...
Thanks Chiqui! Somehow, I never managed to learn it.
What tool do you use to polish a visualization exported from a data visualization tool like Tableau, ggplot, etc. (say using SVG)? Is Illustrator the best solution, or is it something else?
I started using Grok for the first time yesterday. Wow!!! Has anyone tried it? It feels like a completely different kind of dialogue. It's hard to put my finger on what it is exactly, but I am really impressed.
It's fabulous!!! I'll share it with my students. Thanks so much!!!
So timely!
I need your help: do you have examples of visualizations with before and after clutter reduction interventions? I remember seeing a few ones (some animated), but I can't recall where I saw them. Can you send me examples if you have any? Thanks!!! 🙏
Great to hear I was useful somehow!
Are you a "data fiber artist" (representing data through knitting, crocheting, embroidery, sewing, weaving, etc.)? Please help my PhD student Sydney collect information about this fascinating practice for her thesis with this survey: skpurdue.github.io/DataFiberArts/.
Please repost! Thx!
P.s. I have become incredibly skeptical about the relevance of many distinctions we make in visualization. Often, there are bigger problems to address elsewhere in the pipeline.
It seems sensible to avoid gaps when it's a histogram. But overall, I am not sure that it matters a lot. My intuition is that most people have a hard time with histograms, whether there are gaps or not.
IEEE VIS
Being invited to be part of an "important" committee.
This plot is masterful. FT has become the best in news graphics, hands down.
Idea for a new focus app: you can unblock websites/apps only after conversing with a chatbot. If you manage to have a persuasive enough argument, you are granted access.
Start a blog or something similar. Publish intermediate products allowing them to be completely imperfect. Tell yourself you need to publish something every N days no matter what. You can always refine it later. (Side note: morning routines, exercise and cold plunges do wonders for me.)
Yes!!!!!! P.s. "Normal" is boring.
I don’t know if this helps but I feel exactly the same despite all these years. You are not alone! :) I also find that sometimes it’s impossible to please everyone because students want literally opposite things.
For me the bottleneck is not in replying but just removing all the emails I don’t need to reply to and just archive. It’s really surprising there is no solution to that yet. HeyMail has a nice feature to label accounts for being in or out. But that the best I’ve seen so far.
