Director Applied AI Lab & Perry Professor at UC, Author, Podcaster, Tableau Visionary Hall of Fame DataPlusScience.com●BigBookofDashboards.com●ChartChat.live
The sky over the US is emptying.
2,229 passenger planes were in the air over the lower 48 at 3:50 PM ET. At 3:53, the FAA stopped every departure. By 5:08, only 778 were left.
The FAA says it's a telecom outage. The ground stop is still in place. We're watching it.
𝐓𝐡𝐞 𝐃𝐚𝐭𝐚 𝐕𝐢𝐬𝐮𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐒𝐭𝐲𝐥𝐞 𝐆𝐮𝐢𝐝𝐞 𝐁𝐮𝐢𝐥𝐝𝐞𝐫
You can build a custom dataviz style guide for your organization, complete with color palettes, chart examples, and sections on accessibility and best practices.
policyviz.com/2026/04/09/bui…
Forget the Golden Globes, the Emmys and The Oscars, the ONLY award that really matters is the 2026 @Tableau Community Awards, yes folks its time for the 2026 Vizzies! Be the first to cast your votes for the 11th Annual Awards! #tableau#datafambit.ly/thevizzies
Here's another vibe-coded webapp. This one is a live tracker of the NYC Subway trains. I have it set to refresh every 30 seconds using an API, which requires Python (or Node.js), but this is another example of something written entirely by Claude.
This was 10 total prompts. A one-shot instructional prompt launched on my phone. Then 9 follow up prompts to correct various errors over the course of 30 minutes (while doing other things).
I might try a few prompt iterations on this to better show the bread crumbs (maybe lines), and visualizing the current location vs. the next stop vs. future stops.
This year, I’ve been tracing faith through the Bible — studying Scripture, exploring historical context, and connecting themes across time. Reading & Tracing Faith Through the Bible is part reflection, part study, and I’ll keep building on it all year. #datafam#tableau@tableaupublic
Viz: lnkd.in/eu5vdvuP
Plan: lnkd.in/eqRVqgTJ
Yesterday this application didn’t exist. Today it does.
This was 100% vibe-coded. I didn’t write or edit a single line of code.
Before you watch the video, here’s what you’re about to see.
Nothing is sped up. The app is genuinely this fast.
It pulls real-time traffic data for 30 U.S. cities using the HERE API and maps it on a Mapbox map.
The "jam factor" is calculated live and displayed in the "jam bars" and directly on the map.
Then I add five more cities. Each time, the data is instantly added, the bars automatically re-sort, and new points appear on the map.
Across the top, BANs show the average jam factor, the most congested city, the least congested city, and the last update time. All updating dynamically.
Then I pick the custom colors based on the jam factor, customize the color ramp scale, and even modify the bounding box (the area used to calculate traffic conditions) from rural to dense urban settings, or manually define the geography.
How is this done? Knowing what to ask for and iterating with GenAI. I started building this app while I was grading assignments, and just checking back in on the progress!
From zero to a real-time, interactive, API-powered dashboard. That’s where we are.
Insights this morning: Indianapolis had a number of car accidents slowing down traffic, briefly popping it to the top of the list over NYC. But at the moment Miami, FL has the heaviest traffic.
What kind of change will AI bring in the next few years?
@uofcincy Applied AI Lab Director, Jeff Shaffer @DataPlusScience joins me and Donna D next...listen 2:07p 700wlw.com/listen
My annual MRI scan gives me a USB stick with the data, but you need this commercial windows software to open it.
Ran Claude on the stick and asked it to make me a html based viewer tool. This looks... way better.
ICYMI DeepSeek has just unveiled an OCR monster 🤯
DeepSeek-OCR is a 3B-parameter model that redefines document intelligence. It reaches 97% character-level accuracy with 10× input compression, preserving every detail.
Most OCR systems require over 6,000 tokens per page.
DeepSeek-OCR achieves the same with about 100 tokens.
No fidelity loss.
Adaptive modes:
✓ Tiny – simple layouts (≈64 tokens)
✓ Small – standard reports (≈100 tokens)
✓ Database – structured or multi-column docs (≈256 tokens)
✓ Gundam – high-complexity pages (n×100 + 256 tokens)
It scales effortlessly: 200K pages/day on a single NVIDIA A100 GPU.
Its dual-stage design combines a vision encoder for optical compression with a Mixture-of-Experts decoder for structured reconstruction. The outcome is near-perfect precision and zero information loss.
In practice:
➤ 10× lower token costs
➤ Runs on standard hardware
➤ Outputs clean HTML, SMILES, or JSON
While traditional OCRs focus on raw pixel-to-text conversion, DeepSeek-OCR just introduces intelligent semantic parsing 🔥
Open source and free to use.
Repo in 🧵↓