Gavin Masterson, PhD 📊🐍🐸🇿🇦🇸🇪

Conservation biologists be like "I know a spot" then tell you they can't show you because the species is critically endangered and highly sought after in the pet trade.

@grant_masterson @EISAfrica Interesting. I haven't heard of Pinpoint before, so I asked Bard to describe why a media organisation would use Pinpoint. Here is its answer: g.co/bard/share/0cd…

Gingras "Advertising in media is not dead. It's just demanding change. We at Google have for years provided a service called pinpoint which helps media orgs identify who their primary audience is and how to monetize that audience to advertisers." #WAJIC2023.@EISAfrica

@think_or_swim For context, see this thread from @GlobalEcoGuy: twitter.com/GlobalEcoGuy/s…

OMFG. Turns out The Day After Tomorrow could possibly be the Year After Next? We're hanging by a thread from full-scale #climate collapse that will be far, far worse than even our direst imaginings. theguardian.com/environment/20…

@gabbspalomo Marc is such a good guy. One of the nicest people I ever interacted with after I emailed him a question about hierarchical models.

Vamos! @carlosalcaraz just did the hardest thing in tennis. Served for the Championship against the most successful player of all time. Absolutely huge congratulations! May there be many more 💪#Wimbledon

This is rich given that Sweden's birth certificate is basically a printout from a dot matrix printer. I had to get the stupid thing verified with an apostille stamp. 🤪

@PhDemetri @statwonk OK. So then I'm extremely interested in what this "Normal" method is and the validity of the comparison between it and Bayes. Is this in the blog post?

@gavinprm @statwonk I have a blog post draft explaining this, it will come out eventually.

This was a plot I had to make today

@PhDemetri @statwonk To give a little more detail: The conjugate normal prior Eppo uses is expected to produce biased estimates if the assumption of normality is violated. Perhaps you are simply demonstrating that the 'reasonable' conjugate normal prior is not as reasonable as an improper prior 🤷🏽‍♂️

@PhDemetri @statwonk I'm very interested in this discussion. My first question: Have you performed an assessment of the prior predictive distribution? Reason: You mention that you feel the prior is reasonable. This can be scrutinised and may yield useful understanding in a variety of ways.

(🧵) This is the story of how far-right snowflakes afraid of free speech and facts—obsessed, instead, with their Big Important Feelings and using speech to brutalize others—destroyed Twitter. I hope you’ll follow along and share this to the extent the Musk Algorithm allows it.

@PhDemetri @ericmjl It sounds like the data does not have a strong enough signal to 'overwhelm' the prior expectation in your case. To be fair, I have had this experience even with naive priors.

@ericmjl Yea, strong priors.

Do people think Bayesian analyses will "speed up" experiments? I mean, if anything the effect will be regularized and you're going to need more data to overcome good priors.

@BeoMulf_SC2 Finally all my diligent studying of @StarcraftWinter's Low APM challenge is about to make me $5 richer!

?????? tl.net/forum/starcraf…

Louis du Preez & Vincent Carruthers

@MarkScherz Louis DuPreez & Vincent Carruthers.

Does anyone know who took this photo? It has circulated online since at least 2014, but I have not been able to track down who the photographer was, and they deserve credit.

@FattebertJ They must have read my PhD thesis. I hope they cited me.

"many trends in the Living Planet Index need more data to be considered reliable, particularly trends across the global south. In general, bird trends are the most reliable, while reptile and amphibian trends are most in need of additional data"

Hi @StandardBankZA I just got a cold call claiming that you want me to talk to a financial advisor. The person wanted me to set up a Teams meeting with someone. I hung up on them as I assume this is a phishing scam. Can you please confirm this?

@ChrChristensen Without a doubt.

@charliejhadley Totally understand. My devops projects always remind me how much we take for granted once a system is working. To see the experience from the new user's perspective is a very useful (and often humbling) reminder of the pain we have suppressed.

In today's episode of "Things That Shock No-One Other Than @CyrilRamaphosa ..." #Eskom #Loadshedding

@BenCAugustine My brain is slow on a Saturday morning, so I'm not grokking whether the proposed number of changes (n) is selected for each session or an outcome of proposals that are accepted?

Here is what is hopefully a pretty simple explanation of my alternative data augmentation approach for SCR. I struggled with how to classify this algorithm. I think it really should be classified as a form of data augmentation, not RJMCMC, but it has similarities with RJMCMC.