Allen Downey
@allendowney.bsky.social
5K followers 72 following 140 posts
Former professor at Olin College, principal data scientist at PyMC Labs, author of Think Python, and Probably Overthinking It -- blog and book -- and stark raving Bayesian.
Posts Media Videos Starter Packs
allendowney.bsky.social
Sadly, my primary team (the Red Sox) and emergency backup team (the Padres) were both knocked out of the playoffs yesterday.

Now I am left to cheer for my team of last resort (the Notyankees).
allendowney.bsky.social
Sometimes we can use Bayesian methods to infer the effect of selection bias and produce an unbiased estimate.

Here's an example that uses PyMC to solve a classic probability puzzle (the image shows what I think is the original version).

www.allendowney.com/blog/2025/09...
Reposted by Allen Downey
jerthorp.bsky.social
At this point I'm just barely making enough $ on this cohort to cover the platform fees.

It's going to run anyway but I'd really love to have a few more people signed up.

Use the code SIXTY for 60% off at registration.
jerthorp.bsky.social
The autumn evening cohort for Binoculars to Binomials starts Thursday.

Which means you have a little over 48 hours to jump on board!

Get to know your local birds, learn some fun approaches to data viz, and... completely rewire your 🧠?

jerthorp.me/learning
Why birds & data?
To see the whole picture.
Data is a system, not a thing.
To truly understand data, you need to know how it gets made,
how it is changed by the processes of computation, and how
the choices we make in representing it affect the stories that
can be told.
Birding offers a way to get intimate with this entire system. To
watch as our observations become data. To see how they are
recast and reshaped by algorithms and models. To explore the
many practical and poetic ways we might tell our data's story.
Through birding, data is revealed not as static record, but as a
living process of which we (and our feathered neighbors) are
very much a part.
Autumn cohorts of Binoculars to Binomials start September 23rd.
Register now: jerthorp.me/learning Birds of a feather
learn together.
Binoculars to Binomials is an online
course about birding and data
visualization.
Join a cohort of curious people to
build new skills and make
new things.
Autumn cohorts start September 23rd.
Register now: jerthorp.me/learning Why birds & data?
It'll change
vour brain.
> Researchers have shown that cultivating a
birding practice strengthens memory and
improves ability to notice detail. In case that
wasn't enough, being around birds makes you
happier.
Everyday encounters with birds have been
demonstrated to significantly reduce
psychological distress, and improve overall
well-being.
Binoculars to Binomials is an
online course about birding
and data visualization.
Autumn cohorts start September 23rd.
Register now: jerthorp.me/learning
allendowney.bsky.social
I have published five new chapters of Think Linear Algebra!

Read about the project here
allendowney.com/blog/2025/09...

Or jump straight to the book
allendowney.github.io/ThinkLinearA...

And now… Asteroids!
allendowney.bsky.social
Between 2021 and 2024, marijuana was legalized in eight states totaling 18% of the US population. During this time, adult use increased and youth use was unchanged.

Data from NSDUH 2024.
allendowney.bsky.social
As a graduate of an all-boys high school, I am very interested to see the results...
aibm.bsky.social
Two decades ago, the U.S. loosened Title IX to allow single-sex schools.

What does the research tell us about all-boys schools? Ben Smith outlines the highlights:
All-boys schools: What do we know?
All-boys schooling is back in focus as boys fall behind. Explore trends, research, and policies on single-sex education.
aibm.org
allendowney.bsky.social
For anyone who likes LLMs and daytime game shows.
pymc-labs.bsky.social
We made LLMs play The Price Is Right — with real-world CPG prices.

50 models. One goal: bid as close as possible without going over.

🤖 Some nailed it
💸 Some flopped hard

What does this say about LLMs as synthetic consumers?

Full story 👇
dub.link/llms-price-r...
Can LLMs play The Price is Right?
Synthetic consumers—LLMs simulating human survey participants—are becoming a powerful tool for marketing and behavioral research. They promise faster iteration, lower costs, and broader flexibility th...
dub.link
allendowney.bsky.social
Some news articles suggest young men are conservative Republicans with sexist attitudes.

But big picture, young men's views are pretty much on trend.

Here's the data: allendowney.substack.com/p/are-young-...
Are Young Men Veering Right? Not really.
On most issues, gender gaps are small and not much different among young and older people
allendowney.substack.com
allendowney.bsky.social
We're kicking off this new workshop on August 5.
pymc-labs.bsky.social
If you can’t explain your model, can you really trust it?

🎥 Hear from Evan Wimpey on how our Applied Bayesian Modeling Workshop teaches you to build transparent, business-ready models with PyMC.

Seats are limited → dub.link/bayes-modeling

#PyMC #MCMC #DataScience
Reposted by Allen Downey
pymc-labs.bsky.social
Everyone can do ML. Few can do Bayesian modeling.

Set yourself apart with skills that drive deeper insight + better decisions.

Join the next Applied Bayesian Modeling Workshop by PyMC Labs. Hands-on. Practical. Real-world.

📅 Limited seats → dub.link/bayes-workshop
#PyMC
allendowney.bsky.social
I've heard of hard science fiction. This is the first time I've encountered hard economics fiction.
allendowney.bsky.social
ChatGPT and I think that "AI tools are shifting the boundary of what’s viable, and that’s likely to increase the net demand for software development, not decrease it."

More Viable Projects = More Things to Build

chatgpt.com/share/686c2e...
allendowney.bsky.social
Starting August 5, PyMC Labs is running a new workshop on Bayesian modeling in PyMC -- and I'm one of the instructors.

If you want to get started with Bayesian stats, this is your chance! See below for details.
pymc-labs.bsky.social
🚨 We’re opening our private Bayesian modeling workshops to the public — for the first time

After this course, you won’t just build better models.

You’ll know how to explain them — to technical and non-technical stakeholders alike

🎓 Led by top PyMC experts.
📎 Register here: dub.link/68uMjpp
allendowney.bsky.social
I love this -- possibly best ever.
allendowney.bsky.social
Here's what I suspect based on a quick review: chatgpt.com/share/6849c1...

TLDR: Within their subgroups, there's not much variation in gender equality, which is why the observed correlations "disappear"
ChatGPT - Gender Equality Paradox Revisited
Shared via ChatGPT
chatgpt.com
allendowney.bsky.social
Thanks, I am always interested in Simpson's paradox!

The paper sounds interesting, but it has several predictive indicators of bogosity.

If I had access, I could find out.
allendowney.bsky.social
Yes, I think questions like this are good candidates for randomization methods, because you have to specify your model of the system explicitly, which surfaces the hazards of asking "what were the chances?" _after_ you see something surprising.
allendowney.bsky.social
Yes! Using SymPy BlockMatrix to analyze Warren trusses (possibly as an example in Think Linear Algebra)
allendowney.bsky.social
I have a fun story -- can I have a hint about your current location?
Reposted by Allen Downey
pymc-labs.bsky.social
Can LLMs answer open-ended survey questions like real people? 🧠

@allendowney.bsky.social tested GPT-4, Claude & others on 2020 ANES data. Models matched the order of top concerns—but often cited issues (like inflation) that weren’t big then. Useful, but not time-aware.

👉 dub.link/synth-respon...
Can Synthetic Consumers Answer Open-Ended Questions? - PyMC Labs
dub.link
allendowney.bsky.social
Thank you -- it was good to see you!