Peter Gray

Peter Gray

@pgdataviz.bsky.social

Statistical Programmer (Epidemiologist by training) using #Rstats and #Python mostly. #DataScience #Dataviz https://pgdatavizandstats.netlify.app/

73 followers58 following107 posts

I have spent 6 months on a project working with a colleague and manager who just uses AI that, more often than not, won't work or throws errors in production and it's made me utterly miserable I now have my own project where I'm just doing it myself and I'm actually enjoying life again

Libby Heeren@libbyheeren.bsky.social

A widening dividing line appeared this past year between people who used to find joy in their jobs having that joy stripped away by AI, & people who were sort of so-so about their jobs now being wildly jazzed about AI, & I thought I understood what made each camp feel that way, but I was wrong.

Speaking of. I spent hours crafting a Analysis plan for a client, checking, referencing everything. They sent a load of feedback, clearly written by an AI, half of it nonsensical, half of it won't help them at all with what they want to find out... If they can't be bothered, why should I?

I'm currently at the stage "Desperately applying for jobs on LinkedIn during meetings of my current job" If anyone knows what a statistical programmer should do to improve his job prospects in Germany, please do let me know

And the creators of Doom wept, for there were no more worlds left to conquer

coolbutuseless@coolbutuseless.bsky.social

Introducing {rdoom} - Play Doom in #RStats (in time for #UseR2026 !) github.com/coolbutusele... This is a demonstration that {tigerfb} + {nara} can do realtime interactivity in R! Includes: keyboard & mouse control, audio, cheat modes!

The good thing about the internet is that it exposes you to a wide range of opinions, view points, and arguments. The bad thing about the internet is that it exposes you to a wide range of opinions, view points, and arguments.

Saloni@scientificdiscovery.dev

Oh no! Oh no!!

Hello #rstats folks. Any idea when I try to update R to 4.6.0 on Fedora Linux using sudo dnf install R. It says the 4.5.3.1 is already installed. But the latest version is 4.6.0. Do I need to wait for it to be pushed to wherever dnf gets it from? Much obliged

Keep meaning to fully transition over to Positron but now RStudio has the data summary I've stepped further away. I'm just too comfortable with it, it's like a good pair of Boots

In the interest of balance. The UK healthcare system isn't perfect either www.bbc.co.uk/news/article... In the interest of further balance, I could not see this even behind contemplated in Germany

bbc.co.ukFirst debate on NHS single patient record after data concernsReforms to the NHS records system could result in 20,000 fewer visits to A&E annually, according to government projections.
Peter Gray@pgdataviz.bsky.social

After having spent 6 months with German public healthcare data, I've never been more patriotic about the NHS and the data it produces

I'm not saying my workplace is old fashioned but I've just had to spend an hour explaining to my boss (30 years statistical programming experience) git from the beginning (in my second language) and why it's not a "scheiss komplizierte System"

Yeah ok. A couple of things here...

Neuroskeptic@neuroskeptic.bsky.social

"Considering children - a population known to be more cognitively flexible than adults - as a proxy psychedelic group..." pubmed.ncbi.nlm.nih.gov/42190161/ Hmm

Peter Gray reposted

The companies which made the Internet are burning it down. It’s very weird.

TechCrunch@techcrunch.com

Google is transforming Search from a list of links into an AI-powered experience filled with conversational answers, autonomous agents, and interactive interfaces — a shift that could further reduce traffic to publishers across the web.

"We felt there were some solutions in the two different a study plans we asked you to develop within four hours that we wouldn’t have done." Fair enough, but they weren't wrong? Tell you what if you gave me this job and it was developing two study plans every 4 hours, i could see your point

Clients when I tell them this and should adjust their expectations accordingly and not get to granular with their inclusion criteria, but I end up trying to identify patients with recorded metastases on the third Sunday in the month from incomplete claims data anyway.

William B. Fuckley@opinionhaver.bsky.social

It is mind boggling how messy and low quality so much data is. Companies whose entire job is to provide data will turn in shit that barely works.

Peter Gray reposted

"If you don't care about the methods, I don't care about the results"! Most observational research published in biomedical journals is very weak. That's why the methods section is the single most important section, and why it's so ridiculous that Nature journals relegate the methods to the end!

Katie Mack@astrokatie.com

Nice reminder that because a study is published in a journal does not mean it’s a good study or that its stated conclusions are reasonable! When people say they are “doing their own research,” even if reading primary literature, missing the context & subtleties can lead to wildly wrong conclusions.

Authors when you go to the Methodology section to see how they have defined "Ultra processed food"

CNN@cnn.com

Increasing your daily consumption of ultraprocessed foods by 10% — basically the equivalent of a small bag of potato chips — may raise your risk of dementia even if you normally eat a healthy plant-rich diet, according to a new study. https://cnn.it/4n0zNR6

Saturday night ✅ Mid thirties ✅ Reading about the evolution of the pipe in R ✅

Michael Friendly@datavisfriendly.bsky.social

#rstats The idea of a pipe (`|`) operator goes back to early Unix. Eg: `hist | head -5` to see your most recent commands. But, where & how did arrive in R #tidyverse thinking and, importantly: in R syntax? Here's an interesting post that describes this history. adolfoalvarez.cl/blog/2021-09...

Ansgar Wolsing@ansgarw.bsky.social

The Grammar of Graphics for SQL – and it’s for real! 😅 #ggsql brings the Grammar of Graphics to SQL - developed by the team behind ggplot2 at Posit. No need to switch to another tool to create nice charts once you have queried your database. #30DayChartChallenge | Day 22 - New tool #dataviz 📊

The dtplyr package saving my behind a lot as well dtplyr.tidyverse.org

dtplyr.tidyverse.orgData Table Back-End for dplyrProvides a data.table backend for dplyr. The goal of dtplyr is to allow you to write dplyr code that is automatically translated to the equivalent, but usually much faster, data.table code.
Peter Gray@pgdataviz.bsky.social

I'm on the very skeptical side of the AI hype train but, because my boss/colleagues have been asking Claude to build SQL queries forwarding them on to me - that don't actually do what we need them to do and are full of errors- I've had to learn SQL to do it properly. So, it's helped that way #rstats

I'm on the very skeptical side of the AI hype train but, because my boss/colleagues have been asking Claude to build SQL queries forwarding them on to me - that don't actually do what we need them to do and are full of errors- I've had to learn SQL to do it properly. So, it's helped that way #rstats

freeCodeCamp.org@freecodecamp.bsky.social

The R programming language is commonly used for statistical computing & data science. And you can learn it from Harvard in this Intro to R course. You'll use the RStudio IDE and learn about filtering data, writing functions to modularize code, and lots more. www.freecodecamp.org/news/learn-r...

Having spent the last two Working/Learning to code in SQL. May I be forgiven for ever complaining about an Error message ever unclear. May as well write them in Sanskrit

The other 20% is explaining why projects are behind because I'm debugging my colleagues scripts that they just fed an SAP into an AI and asked it to generate the R script for them...

Ben Harrap@bharrap.bsky.social

On one hand, I could use packages like renv to mitigate versioning issues On the other hand, I could just update every package each time I open R and fix whatever breaks #rstats

A very disheartening aspect of my job is, when asking experienced colleagues for help with my analysis or debugging , is they'll either tell me to "chat with the AI" or send me a link to a chat they've been having with an AI. Not exactly conducive to my learning. #rstats

Peter Gray reposted

OpenSAFELY is open from today! Huge thanks to all who supported this vast collaboration: whole population GP data; in a productive platform; innovative privacy protections; unprecedented support from professions, privacy campaigners; &c Now it's over to users! www.bennett.ox.ac.uk/blog/2026/02...

bennett.ox.ac.ukOpenSAFELY news: you can apply to do non-COVID research, from today! | Bennett Institute for Applied Data ScienceWe are delighted to announce that - from today - you can submit applications to the OpenSAFELY service for non-COVID-19 studies.