Zach Nabavian

Ugh that sucks. I hope you feel better soon.

*That

Thank thing is huge 👀

I don’t drink coffee at all, but this mug is looks great, and I love that Steven Universe mugs are out there.

IRRC this was Julia? Daily trivia sounds really fun.

It’s fun to see a smiling pumpkin

That one on the left is simple???

I think it is supposed to come to NSO soon!

Also I never got to play a ton of Double Dash. I really should at some point.

I’m trying a bunch of different characters in Mario Kart World. I tend to pick Yoshi a lot though. My favorite one is usually whatever is most recent, but I really like Mario Kart DS.

Mentioned this before on the data science hangout I believe, but I’m obsessed with Mario Kart.

I don’t really think about commenting my code that much. A lot of the time I think “this should be pretty self-explanatory” though my code is also relatively simple. I’m not really a software engineer. I just like R for data manipulation and plotting.

I agree. This is pretty big reason why I don’t use Bluesky as much as I thought I would.

Employment Situation Summary - 2025 M07 Results

U.S. employers added 22,000 jobs in August and the unemployment rate ticked up to 4.3 percent.
Data: www.bls.gov/news.release...
Live coverage: www.nytimes.com/live/2025/09... #NumbersDay

Weirdly enough, I have the same experience when people say “exact” or “exactly”. I always hear it like “Ec-ZACH-t”

That must be really confusing lol.

But that difference could just be due to the limitations of the model and noise in the data set. I do intuitively think that moderate candidates are likely to do better, but WAR models don’t seem like the best way to show that, unless I’m completely misreading them.

I might be misunderstanding these WAR models, but they seem to me like they first build a model that accounts for fundraising, partisan lean, incumbency, etc, and then attribute the difference between the model and actual results to the strength of the candidates.

This may or may not be because of AI, but I notice that a lot of LinkedIn posts follow the same general structure and tone, probably to optimize for engagement.

Added salt is seriously the biggest trap ever. Why did we evolve to like it so much???

No one hates things more than fans of those things.

Yeah, following instead of connecting feels more appropriate for people outside of the immediate network.

LinkedIn is frustrating in that unless you have Premium, you can only send a few personalized invites a month. Ideally they would be free for everyone. Ideally personalized invites would be free and everyone would add a note when they wanted to connect.

Fixing code would actually be fun if was more like a Mario game.

Ngl when I first came across this I thought of SpongeBob and Patrick going into the treedome for the first time and being unable to breathe air.

My afternoon slump has always been frustrating, but this is so much worse.

On a more serious note, ggplot2 is my favorite visualization tool by far.

Wow, now ggplot2 can go off to college.

Also, there is a potential limitation to this dataset. There was only 1 earthquake recorded in each of 2011 and 2012, while hundreds were recorded in subsequent years. This makes me wonder if the dataset is incomplete. I don't think it's likely that those early years had one earthquake each.

GitHub - Zach-Nabavian/tidytuesday Contribute to Zach-Nabavian/tidytuesday development by creating an account on GitHub.

It was fun to experiment with this kind of graph, anyway. Here's my github: github.com/Zach-Nabavia... Thanks to @libbyheeren.bsky.social for curating this week's dataset.