PyMC Developers https://bayes.club/@pymc

@pierreanalytics @pymc_labs I actually did have a venue effect in one version of the model, but decided to keep things simple (maybe too simple). Once you are into more than one season of shot data you'd want to be using a GPU or switching to approximate Bayesian methods.

@pymc_labs Jason Dickinson being a good example (his 23-24) - Curious on your thoughts on creating a model based on multiple seasons in this case where we are seeing pucks tracked differently. Or things like rink bias in general x.com/pierreanalytic…

Hockey goaltending stats are noisy. We used Bayesian spatial models to separate skill from luck, creating a fairer framework for scouting. See how Gaussian Processes clarify performance and improve decision-making. 👉 dub.link/xWXqutU #SportsAnalytics #PyMC #Bayesian

@32NYR @pierreanalytics @pymc_labs Yes, this is just one year. Also keep in mind that (1) those effects are just for wrist shots and (2) the width of the credible intervals mean he could actually be much lower than top-5. Multiple years would definitely improve the shooter random effect estimates!

@pierreanalytics @pymc_labs has to be a single-year (2023-24) no? given that Carcone is 2nd in top-5 shooters based on estimated effect

@pierreanalytics @pymc_labs Thanks Neil! You would definitely want a temporal aspect to the model; the post was really just a demonstration of continual model expansion as opposed to a finished product. Spatio-temporal HSGPs work great, though at some point we'd probably have to switch from MCMC to VI.

@pymc_labs Great read, I have a couple questions: - I really like the approach of incorporating shooter quality into a GSAx model, did you consider multiple years for data or some priors? Shooting can vary so much over a year but may not be representative of the difficulty of their shot

🚀 Big news for Mac users! PyTensor now supports the MLX backend - bringing Apple Silicon (M1/M2/M3...) hardware acceleration to PyMC. ⚡ Try it out by setting mode="MLX" and share your feedback! Huge thanks to all contributors! 🙌 #PyMC #PyTensor #MLX #AppleSilicon

📢 PyMC Office Hours - You're Invited! Got questions about Bayesian modeling or PyMC? Join us for an open chat with Jesse & Christian - bring your projects, ideas, or just curiosity! 📅 Oct 20, 2025 — 8 PM NYC | 7 PM Chicago 📍 Zoom (RSVP 👉meetup.com/pymc-online-me…)

State-space modeling in PyMC just leveled up! 📈 We've released Dynamic Factor Models in pymc-extras v0.5.0, achieving 1:1 feature parity with statsmodels. A massive contribution from GSoC 2025 student Andre Cate. Get started with the notebook: github.com/pymc-devs/pymc…

🚀 New in PyMC (dev): pymc.dims - build PyMC models with named dimensions - Auto alignment & broadcasting by named dims - Fewer reshapes, transposes & hacks - Use softmax(dim="item") instead of axis=-1 Read more: #dims-module" target="_blank" rel="nofollow noopener">pymc.io/projects/docs/…

PyMC 5.21.0+ is here with support for NumPy 2.x & Python 3.13. If you held off on upgrades, you're clear to go! 🚀 Details 👉 github.com/pymc-devs/pymc… Big thanks to everyone who made the migration smooth! #pymc #numpy #python #opensource

Excited to welcome Open Wound Research as a sponsor of @pymc_devs! 🎉 They're advancing wound care research with open science, open-source tools, and Bayesian modeling — powered by PyMC Learn more about them:openwoundresearch.com Follow their journey:linkedin.com/company/open-w…

📍 Join us for the PyMC Code Sprint at #PyDataLondon 🗓️ June 7 | 🕥 10:20–12:35 🔗 cfp.pydata.org/london2025/tal… Contribute to #PyMC—fix bugs, write docs, build examples—and learn directly from core devs. No prior experience needed—just your laptop and curiosity! See you there!

🎙️ PyMC Office Hours — May 14! Come chat about Bayesian modeling, PyMC, or just hang with the community. Whether you're a beginner or an expert, all are welcome! 🗓️ Wed, May 14 | 🕒UTC 13:00 | NY 9am | LDN 2pm 📍 Zoom (RSVP for link): meetup.com/pymc-online-me… #PyMC #OfficeHours

PyMC is in Google Summer of Code 2025! We're excited to be part of #GSoC2025 under @NumFOCUS! If you're passionate about #Bayesian stats & #OpenSource, this is your chance to contribute to #PyMC! 📅 Deadline: April 8, 18:00 UTC 🔗 Apply now: pymc.io/blog/blog_gsoc…

📢PyMC Office Hours! Got questions about #bayesian stats, #modeling tips, or just want to chat? Join us on Monday, Jan 27th, for a relaxed session with the PyMC community! Everyone's welcome. 🤝 Link: meetup.com/pymc-online-me… 📅 Date: Jan 27, 2025 | 15:00 UTC 🕒 Duration: 1 hour

👋 Hey Open Source Enthusiasts! Join us for the PyMC Sprint at @PyData NYC 2024! We are excited to be mentoring this sprint and can’t wait to support you in making contributions to PyMC! 🗓 Event Details: meetup.com/pymc-online-me… Let's collaborate & learn with @NumFOCUS @PyData

I've been wanting to blog for a while. Today I finished this post about data simulation and parameter recovery with PyMC. It covers the do and observe operators, as well as the classic coords, dims, etc. tomicapretto.com/posts/2024-11-…

Hanging out at the PyMC office hour right now! Drop in if you’ve got questions or just want to chat about Bayesian stuff. numfocus-org.zoom.us/j/86159816171?… #PyMC #OfficeHours #ChillChat

📢 1 day to go! Tomorrow, we’ll be hosting another exciting round of office hours where you can bring all your questions about PyMC. Whether you're troubleshooting a problem, curious about Bayesian stats, or just want to hang out and chat, we're here for you. See you tomorrow!🤝

Hi PyMC community 👋 We are holding office hours next week. The office hours are open to everyone, and anyone should feel welcome to attend. 📅 𝐃𝐚𝐭𝐞: Wednesday, 16th Oct, 2024 ⏰ 𝐓𝐢𝐦𝐞: 14 UTC / 7 am PT / 10 am ET 👉 RSVP: meetup.com/pymc-online-me…

🔥PyMC 5.13.0 is here!🔥 - Trace too big? Use var_names to only store those you actually need. Need them later on? Use compute_deterministics() - Truncate custom distributions with CustomDist for flexible modeling ⚡ Python 3.10+ support, and more: github.com/pymc-devs/pymc…

🚀 PyMC 5.10.4 has been officially landed on Colab! This latest version brings seamless integration of the numpyro sampler. Now, you can run blackjax and numpyro smoothly. Explore all the features and enhancements; check it out now on Colab: colab.research.google.com #PyMC