I hate on MS Teams a lot but their device transfer (moving you from a call on a phone to your PC) is flawless

The #rstats community, generally and specifically at conf, is THE number one reason why I keep using R. It’s like nothing I’ve experienced with other languages I’ve used, and it is POWERFUL.

My first R class had ~100 students. For a web scraping lesson, homework was to scrape job sites for posts, compile data of interest, and analyze it. But the prof forgot to teach rate limiting, so just before the due date the entire class basically DDoSed and brought down the major job sites #RStats

RIP to anyone named Al (lowercase L) in these trying times #AI

Learning projections will ruin you for life

I think I need a year's sabbatical just to practice and internalize all the developments in R that have taken place since I went back into management :( #RStats

I've seen tragically few Robert Redford movies, but "Sneakers" is truly underappreciated. Its almost a period piece of post-cold-war paranoia, with an exceptionally prescient perspective on the surveillance state and burgeoning concerns over information security. Plus the cast is 💯 highly recommend

massive gen AI win just now at $DAYJOB -- I gave GPT-5 a long list of names of people whose internal email addresses I needed. It came back with the exact structure I requested and provided additional details about its assumptions when matching names to addresses. Saved me like an hour. #genAI

Time for a Friday #dataScience #canOfWorms! Agree or disagree: data science deliverables (i.e. outputs for the business) are digital products, and should be designed, developed, and governed as such.

putting my corporate windows machine over the A/C vent so the fan turns off before I put the computer to sleep 🤡 #notSatire

"When offices are positioned and built to draw people in, then community develops." As a member of a fluorescent-lit, cubicle-rowed, windowless corner of an office, I've been talking myself hoarse about the incentives that make a workplace a draw. www.businessinsider.com/cbre-announc...

as a lowly econ master's student with no stats training, I asked my advisor what fixed effects were. He hesitated. "With fixed effects," he began, "well... the effects are _fixed_". It was a strange moment.

Obligatory commitstrip (from the pre-gen-AI era!) www.commitstrip.com/en/2016/08/2...

I had a visceral stress reaction to this image

Do not make your PhD your identity. Yes, you pursue your studies because of your passion. But give yourself the freedom of objectivity and have a backup plan so that you can walk away confidently if it comes to that. It's a big world out there, and there are so many ways you fit into it.

I'm not listening to James Horner's Apollo 13 theme on repeat, YOU'RE listening to James Horner's Apollo 13 theme on repeat

It's Friday, let's open a can of worms! To what extent do principles of software development apply to #datascience? Not a novel question, but still a sincere (and intentionally broad) one. DS practice is so varied, and I'd like to collect perspectives on when DS work needs to consider the SDLC.

how to make a GUI work for you, rather than the other way around:

I plan to post a few thoughts about the Pixel 10 after more time using it. But today someone on the phone asked for the number of one of my contacts and I was certain Magic Cue would dig it up right away. Reader, it did not. #pixel10 #google

Every time I see that last photo, I think that conf is at Hogwarts this year. maybe next time

python users at $DAYJOB needed help running some legacy R code. so I set up a meeting called "attend this meeting if you love R" and I'm genuinely wondering if they'll show up #rstats

there's a metaphor about python environment management buried in here but I haven't fully worked it out yet

oh hi! I know what you're thinking, insurance sounds super boring. trust me it's not, and we have the panelists to prove it! ⤵️

Blade Runner: the future is dystopian and humorlessly bleak Reality: the future is dystopian but at least it's a comical self-parody

Hey #rstats do your thing! 🙏

the moral of the tortoise and the hare is "slow and steady wins the race" which always bothered me because it's predicated on your very competent opponent taking a nap by the finish line

Y'all, I waited three months for this book to come in at the library. Cheers.

Writing is thinking. Business leaders value a volume of words over volume of thought. Therefore LLMs will be pushed by business leaders onto people whose job is knowledge work. There's some kind of syllogism buried in there, but there's your half-baked idea for the day.

telling that people who are professionally Very Computer so often want Less Computer in their lives

I think this makes me a boomer but I just started using my laptop to send SMS messages and it's so much nicer

When doing #datascience in business, there's a tension between org standards and team innovation. E.g. in the of collaboration I've heard "we all manage environments this way" or "this is our modeling package". How do you manage the tension and promote new ideas where rigidity is generally favored?

let's make #LLMs *really* useful and add error bars to every token of their output

I have been experimenting with different ways of visualizing the curvature of meandering rivers. Plotting curvature in the context of the meanders themselves helps QC-ing the curvature data ⚒️🧪

Just a lil reminder about the handsomest panel you ever did see! Come hang next Tuesday! #rstats #datascience #posit

terrific scrollytelling technique to drive home responsible data visualization practices. I'll be sharing this with colleagues. oh and it's built in quarto using #rstats and the closeread extension! 💯

look everyone let's just settle this thing #chatGPT

As a little experiment, over the summer I've been using the Gemini chat on my Pixel phone as my personal mixologist. I've really enjoyed almost all of its recommendations! Here's a lil thread with some thoughts on the interaction with the AI >> #genAI

it's not an exaggeration to say that most of my troubles managing *kids these days* in data science comes down to modern DS programs don't teach this way

hot take of the day is that if you're a python user, you're not allowed to complain about people using two spaces after a period. v sorry I don't make the rules

now I have to tell my attorney to put a non-resurrection clause in my will and I am extremely annoyed

it finally happened. an amazing job candidate turned us down strictly on the basis of the (company-mandated) hybrid work schedule

coworker said "wouldn't it be unsettling to see a perfectly straight banana?" so of course I'm trying to make gen AI create one. but like the "full glass of wine" problem, the AI cannot give me a banana without a curve #genAI

if bluesky had launched in the golden age of the internet circa 2004

Flashback to when, as a junior statistician on a team of otherwise brilliant scientists, I would pull my hair out at my desk because a half hour consultation before designing their experiment would have saved me hours of garbage analysis

as I begin to leave behind "ordinary" ML and take on more responsibility over agentic AI projects (gag), be prepared to hear me complain constantly about the use of the term inferencing to mean predicting/scoring

whitespace as syntax -> never!

So I guess there will come a day when I have to get Lasik just so people know I'm not using AI glasses to record everything and to feed me conversation tips

24-year-old me would have *killed* for this kind of resource while trying to learn LOESS from the primary lit grad school

inject this directly into my veins plz