iiinova — where your digital self keeps evolving

Tomi Capretto @tomicapretto.bsky.social · 02/07/2026

The PyMC blog isn’t known for being super active, but an excellent post summarizing all the great new stuff in PyMC 6 and PyTensor 3 landed a while ago. Enthusiastic PyMC users, this is a must-read: www.pymc.io/blog/pymc_v6...

pymc.io

PyMC 6.0 & PyTensor 3.0: ecosystem updates

PyMC has been under steady development since the early 2010s. To mark the new major releases of PyMC 6.0 and PyTensor 3.0, we want to highlight the developments across the PyMC ecosystem(and its cl...

030

Reposted by Tomi Capretto

Aki Vehtari @avehtari.bsky.social · 26/06/2026

@aloctavodia.bsky.social has been porting Bayesian Workflow book avehtari.github.io/Bayesian-Wor... case studies to Python using CmdStanPy, numpyro, ArviZ, Bambi, and kulprit arviz-devs.github.io/bayesian-wor... He is working on porting the rest of the case studies, too

arviz-devs.github.io

Bayesian Workflow book: Case studies in Python – Bayesian Workflow case studies in Python

16218

Reposted by Tomi Capretto

Aki Vehtari @avehtari.bsky.social · 23/06/2026

New paper "To select or not to select: predictively consistent priors instead of model selection" with Anna Elisabeth Riha, Leevi Lindgren, @davidkohns.bsky.social, @paulbuerkner.com arxiv.org/abs/2606.22850 Model selection is not a substitute for building good models in the first place 1/

To select or not to select: predictively
consistent priors instead of model selection

Anna Elisabeth Riha, Leevi Lindgren, David Kohns, Paul-Christian Bürkner, Aki Vehtari

Bayesian modelling workflows often consider multiple candidate models of varying complexity. Model selection is commonly used to navigate potential trade-offs between model complexity and generalisability to new data. We study when model selection is unnecessary or can even be harmful for predictive performance in finite data regimes and find that the need for selecting simpler models can depend on prior choice. We formalise predictively consistent priors, which keep prior predictive implications stable as model complexity increases. Across examples and numerical experiments, including adding covariates in linear and logistic regression, forward variable selection, and nonlinear modelling, flexible models with predictively consistent priors typically match or outperform selected simpler models in out-of-sample predictive performance. When selection helps, it can indicate poor joint prior implications, such as excessive prior mass on implausible predictive values. Based on our findings, we propose replacing the notion of sparsity or parsimony at the level of model components with specifying priors that remain sensible in predictive space as models become more complex.

110631

Tomi Capretto @tomicapretto.bsky.social · 21/06/2026

I don't understand Slack. I opened it after months and, as expected, I had to log in. What I didn't expect was that I had to log in to every workspace, all using the same e-mail address. It seems that was not enough, as Slack decided I had to copy/paste a security code in every log in attempt.

010

Reposted by Tomi Capretto

Chelsea Parlett @chelseaparlett.bsky.social · 12/06/2026

The goal should never be "use ai ", the goal is "solve X problem" and sometimes ai could be a tool to do that

4304

Reposted by Tomi Capretto

ArviZ @arviz.bsky.social · 07/04/2026

ArviZ now has built-in tools for prior & likelihood sensitivity analysis via power-scaling! Instead of fitting multiple models with different priors, you fit once and use importance sampling to approximate the effect of perturbing the prior or likelihood.

1167

Reposted by Tomi Capretto

ArviZ @arviz.bsky.social · 01/04/2026

ArviZ 1.0 brings a lot of new plotting functionality. Check our gallery for a glimpse of what's new: python.arviz.org/projects/plo...

python.arviz.org

Example gallery — arviz-plots dev documentation

195

Reposted by Tomi Capretto

ArviZ @arviz.bsky.social · 19/03/2026

ArviZ 1.0 is out! We have refactored it to be more modular, flexible & lightweight. For an overview of the changes, check the migration guide. python.arviz.org/en/stable/us...

python.arviz.org

Getting started — ArviZ 1.0.0 documentation

21510

Tomi Capretto @tomicapretto.bsky.social · 20/01/2026

It took my just 7 years of Python coding to embrace the power of `breakpoint()`. Feels so great!

000

Reposted by Tomi Capretto

Alexandre Andorra @alex-andorra.bsky.social · 02/01/2026

This one is a foundational one: #ExperimentalTrialDesign the ̶r̶i̶g̶h̶t̶ Bayesian way, with no other than Scott Berry, from Berry Consultants, LLC! Enjoy, and happy new year, my dear Bayesians 🖖

052

Reposted by Tomi Capretto

Alexandre Andorra @alex-andorra.bsky.social · 02/01/2026

My #AdvancedRegressionModeling course, written with the brilliant Ravin Kumar and @tomicapretto.bsky.social, is now available through my Topmate profile! So do give it a try and let me know what you think in the comments 👇 See you soon in the Intuitive Bayes' Discourse 🖖 topmate.io/alex_andorra...

topmate.io

Your All-in-One Creator Storefront

Make money from your content. Sell products, host sessions, and grow your business — all from a single link.

032

Reposted by Tomi Capretto

The R Foundation @r-foundation.bsky.social · 10/12/2025

The R Foundation is happy to be a partner on the project "Enabling the next generation of R contributors" awarded £500k under the Research Software Maintenance Fund (Software Sustainability Institute, UK). Announcement: www.software.ac.uk/ssi-awards-f... Congratulations to the project team!

software.ac.uk

SSI awards funding to 13 projects through the RSMF Round 1 | Software Sustainability Institute

We are pleased to announce that the Research Software Maintenance Fund (RSMF) has funded 13 projects to advance the sustainability and impact of research software. Through this initiative, the SSI, wi...

14315

Reposted by Tomi Capretto

Stefano @stippe87.bsky.social · 07/10/2025

Modelling nested factors in Bambi+PyMC #stats thestippe.github.io/statistics/n...

thestippe.github.io

Nested factor

Saving yourself from mis-labelling

051

Tomi Capretto @tomicapretto.bsky.social · 26/09/2025

If you had to bet against a free throw, would you bet against the first or the second attempt? What if the player were a rookie? And what if it were a specific rookie? Let's try to find out. tomicapretto.com/posts/2025-0...

010

Tomi Capretto @tomicapretto.bsky.social · 09/08/2025

TIL in Python about any("") all("") any([]) all([]) 🤯

000

Tomi Capretto @tomicapretto.bsky.social · 24/07/2025

Attempted to execute a cell in a Jupyter Notebook in Positron to see what the dependency installation process was like... It just worked, no extra dependencies, just Python. Python data science tooling is healing.

010

Tomi Capretto @tomicapretto.bsky.social · 07/07/2025

Empecé con R/RStudio y fue un bajón cambiar de IDE para Python (terminé en VS Code). Hace poco salió la 1° versión estable de Positron, que mezcla lo mejor de RStudio y VS Code. Como pronto arranco a dar un curso en Python, armé tutorial de instalación: www.youtube.com/watch?v=9orV...

youtube.com

Descargar, instalar y usar Python y Positron en Windows

YouTube video by Tomas Capretto

010

Tomi Capretto @tomicapretto.bsky.social · 06/06/2025

Polars is so beautiful I can't look at pandas data processing without my eyes bleeding

020

Reposted by Tomi Capretto

Mine Doğucu @minedogucu.com · 05/04/2025

Please help us spread the word to teachers about StatisticsTeacher.org @amstatnews.bsky.social @nctm.org #statsed #nctm #math #statistics

Text of the image reads “You can find articles and free lesson plans specifically for teaching statistics and data science in K-12. Check out StatisticsTeacher.org. Consider submitting your own article and lesson plan.

0208

Reposted by Tomi Capretto

Allen Downey @allendowney.bsky.social · 26/03/2025

At PyMC Labs I've been working with a group developing synthetic consumers for marketing research. We just published this white paper with an overview of work in this space -- and we have a blog post coming next week with some experimental results. www.pymc-labs.com/blog-posts/s...

pymc-labs.com

Synthetic Consumers: The Promise, The Reality, and The Future - PyMC Labs

072

Reposted by Tomi Capretto

Nathaniel Forde @nathanielforde.bsky.social · 24/01/2025

Looking forward to this! Talking about the do-operator and structural causal models in PyMC

3124

Reposted by Tomi Capretto

Alexandre Andorra @alex-andorra.bsky.social · 24/01/2025

Really fantastic to host Jesse Grabowski on the show! I learned so much about #TimeSeries thanks to him. And there is still so much I don't know, so I guess you'll have to come back very soon Jesse ;)

072

Reposted by Tomi Capretto

Alexandre Andorra @alex-andorra.bsky.social · 21/01/2025

🚨 **Big News!** 🚨 The first two lessons of our Intuitive Bayes courses are now FREE! 🎉 Try them out to make sure they're the perfect fit before diving into the full experience

164

Reposted by Tomi Capretto

Pierre-Simon Laplace @learnbayesstats.bsky.social · 11/01/2025

💯What if Bayesian modelling could be faster, more flexible, and easier to interpret? 🎧 youtu.be/7POdNknJ1Es?... 🎙️ Episode 123 is here! @alex-andorra.bsky.social chats with @aloctavodia.bsky.social about ground breaking tools and ideas that are shaping the future of Bayesian workflows. #LBS #BART

0175

Reposted by Tomi Capretto

Alexandre Andorra @alex-andorra.bsky.social · 10/12/2024

Max Göbel and I recently submitted a paper to the MIT Sloan Conference. We're using #HSGPs to isolate ⚽ players' contribution from team effects and introduce Skill- and Performance Above Replacement metrics. Paper is still WIP, but we put it on Arvix to get first feedback! arxiv.org/abs/2412.05911

084