My daughter has no idea the pain she is standing on.

Best use of NFTs.

I don’t know what’s going on. I don’t know technology. I don’t know what AI is. I don’t think you get to know what AI is and also see a donkey jump off a high dive. I think it was one or the other, and I saw the donkey, so I’m out.

TIL about jeep ducking and I’m comforted to know I don’t just keep seeing the same jeep all over town

KPop Demon Hunters is modern day Wizard of Oz I said it.

You should use AI while coding. You should also understand every piece of code you commit. Try to at least. Not everyone seems to agree. So here are 2 obvious signs that someone has mindlessly passed along Copilot generated python code: - consistent, sentenced capitalized code comments - filtered_df

That’s tough to overcome

I’ve tried convincing myself for the past 2 years that pyspark doesn’t stink. But it’s the worst. I can’t wait for something better to come along.

If you’re code requires .cache() to work properly, either your code stinks or the framework you’re using stinks.

Did I call my work AI at the time? No. Am I glad it gets called out as AI now? Yes. Should people have realized the value of this work years ago. Also yes. Looking at you, people I ultimately still like but wish things could have worked out differently 👀🐅 doi.org/10.1080/1058...

🤤

An issue (feature request) I opened for RStudio 7 years ago was recently fulfilled and closed. Amazing. github.com/rstudio/rstu...

Finally after all this time I’ve finally found my way to GitHub as a part of my actual job.

Synchronous = 23.5 hours Async + parallel = 20 mins

I *only* present slides that I want to drain.

Pyspark where your code can be correct but because of weird spark things you must include a .cache() else you will leak data in each fold.

The Data Science moment has indeed ended. Thank you for the past ten years, @rdpeng.org and @hspter.bsky.social!

My 4-year-old daughter just wrote a note to her 5-year-old cousin. I helped her spell her cousin’s name. A couple of minutes later, she asked how to spell “Like and subscribe.”

They call it Staples because they use staples to repackage your Amazon returns.

My daughter thinks it’s hilarious that I referred to my phone’s background image as a “wallpaper.”

Being an adult means appreciating that Kevin McCallister’s mom is one of the most talented people alive.

Our CTO frequently says, “Feedback is a gift, y’all,” and it is one of my favorite things.

This law firm thing seems especially clear cut (i.e., indefensible) and wild.

This project is so creative/simple/brilliant.

There are two kinds of data scientists. Those who have owned production code running daily for over a year and those who have not.

Here’s a pet peeve: 👉 Using “simplistic” (oversimplified, reductive) when you actually mean “simple” (clear, straightforward). Example: it’s less complicated than we thought, so let’s implement a simplistic solution and move on!

Watching videos on how to actually use PowerBI, I can feel data science brain cells leaving my head.

What color palette is this?

On Microsoft Teams, how often should people be able to use the [@]Everyone tag? Once a day? Once every 100 messages?

Most people think a "millennial" is someone born in a specific range of years. That’s incorrect. A millennial is actually someone who uses numbers in their SQL group by statements, excessive whitespace, VLOOKUPs, Firefox, your name every time they talk to you, etc.

My career.

It's amazing the things one can do in base R (without installing or loading any other #rstats packages):

One of the cooler projects I’ve never got back to. github.com/mkearney/wac...

✅🎯🏆

Leadership: Data need to move to Azure so this KPI needs to switch over as well. Data Scientists / Engineers: Is the KPI important? How accurate do we need to be? How much time do we have? Leadership: People look at it every day at the basis point level. And ASAP. Data Scientists / Engineers:

Can’t imagine why support for this baby camera was discontinued.

Also, shameless self-promotion-- I'm a major R and data nerd, and I'll be on sabbatical next year. I would love to get involved with journalism projects (either paid or unpaid). If you might have something I could help with, come say hi!

I want to be able to @{entity} when prompting AI. Instead of having a static context, I want to be able to easily populate it. @{work} What are the names and ages of @{person}’s kids? I’m 158 yds from hole which club in @{mygolfbag} should I hit?

I’ve come to really appreciate the named tuple in python. Still not the most elegant implementation but it has just about all the behaviors I want without the baggage of dataclass, pydantic BaseModel, etc.

My first #NICAR2025 conf is right around the corner & there are 213+ sessions! 🤩📊 If you're like me & prefer a calendar view, I made a tool just for you! It's an #RShiny app that lets you search & select sessions to export as an .ics calendar file GIve it a try! tiny.cc/nicar-2025 #RStats #RLadies

turns out ~50 domains had 14 or more severe flails since I started monitoring (~month-ish now). That doesn't seem super "efficient". We shld prbly disband DOGE. 🐝swarm plot; Y axis is # flails In other news, 🇺🇸 are so daft the Capitol Gift Shop had to register the `a` version of their name.

“Value” lost on meme coins since Jan: $50 billion Actual money lost on meme coins since Jan: $10 million AI thinks that 2nd number is low (I think it’s high), but just bc a coin app says your meme coins are worth $100,000.00 doesn’t mean you can actually sell for that much. I have zero sympathy.

Idk when it happened but at some point I made a 180° on the clickable “Copy” button. I *hate* dragging the cursor around text I want to copy. It feels like Windows 95.

The Best Way to Use Text Embeddings Portably Is with Parquet and Polars Discussion

Like most, I’ve joked about it raining while sunny, secretly knowing there was a cloud nearby that I just couldn’t see. Then I moved to Michigan and experienced an hour of snow without a single cloud in this whole f***ing chunk of atmosphere.

I’ve come up with a way to measure dominance, momentum, and resilience and they are all factors in my network-based NCAA men’s basketball ratings.

Something is wrong with Google News today. It thinks it’s the 1950s.

This Kansas team is so bad.

I used eigenvector centrality to rate men's CBB teams by building a directed network where losses (loser→winner) are weighted by log-transformed, clipped MOVs and adjusted for location (.75 home, 1.0 neutral, 1.25 away), rewarding wins over strong opponents. Below: top 30 & deviations from KenPom.

Top 25 men’s college basketball teams according to eigenvector centrality weighted by ratio of victory. Purdue #2. Duke #25. Interesting!