Nathan Jeffery

Many of my colleagues have limited Shiny experience, and I wanted a debugging option that was as simple and legible as possible.

I use this in a shiny app! 'debug' is a list containing items for whether to 'write_rds' files to a debug folder after major data cleaning steps and/or 'print' debug messages (to the console and/or a .txt file, using the logger pkg)

If it's a data frame in a function, it's called `dat`.

If it's a data frame I'm prototyping something with, it might also be called `dat`. Or `foo` or `bar`.

Yeah, this feels like a job for brand.yml and some nice code templates

I loved hearing your talk today. What a useful and thoughtful package. Thank you for sharing it with us!

I need you to know that your contribution, your voice, your perspective, your way of explaining something, YOUR whatever, is valuable.

Just because "it's been done before" doesn't mean you shouldn't do it. Nothing is new, everything's been done! But not by you. Yet.

#databs #rststs #python

Please stop spamming #rstats with your bots

Use a descriptive tag so we can can choose to follow, and not have to trawl through so much churn

I can't see any option other than blocking the accounts, I'm interested in CRAN updates (and others) but gee, it should be about humans and actual ideas

Growing up, my grandma had an oblong kitchen table. Wasn't til today that I looked it up though!

Apparently "oblong" is just generic for "a stretched out square or circle"?

IDK

Does some of this functionality depend on Shiny version 1.11 to update if module files are changed?

Hey, I'm sorry. My bad for misunderstanding you.

Anyone an idea where I can find the source code of ardata.fr/ggiraph-book/ ? The github link on the website is dead. #ggplot2 #rstats #ggiraph. Thanks!

Pak is so fast!

I agree that aesthetically it looks better and highlights that the foreground data is smaller, but I think for making sense of the graph, the only justifiable options are having the foreground bins the same width or half the width of the background

And the more we discuss the humanness of intelligence, the ways we can see and cultivate it,

The more it becomes blindingly obvious that (a) we deserve better out of our education system, and (b) AI isn't all it's cracked up to be.

end/

Would I go to college again?
Yes.

But I've seen enough blindingly intelligent people without PhDs (and a handful of "meh" people with them) to know that "PhD level intelligence" ... isn't entirely a function of having a PhD.

6/

Should we encourage people to learn to write more clearly? Yes.

To read? Definitely.
To collaborate? Absolutely.
To develop a passion? Of course.

Is college or a PhD the best place to do that?

I'm not sure. It's a good one, but it's not the dream I was led to believe it would be.

5/

And frankly, dissertations suck. If you wrote one, you should definitely be proud of getting through the research, the red tape, the revisions, the pages and pages of writing. It's a feat.

But I don't delude myself that I would be more intelligent if I had written one.

4/

Do I think better, write more clearly, know more facts, and have more empathy than when I started my PhD? Yes.

Did school help develop those things? Absolutely.

But a PhD is only one way to structure the effort, rigor, care, and feedback that lead to intelligence (however we define it)

3/

dichotomy of "intelligence as regurgitating facts" vs "intelligence as an ability honed in college, especially during a PhD."

As someone who just left a PhD program (and works with lots of folks with PhDs): we can't be holding the PhD up as the pinnacle of intelligence.

2/

I agree with a lot of this, especially that the AI bros calling it "PhD level intelligence" shows a fundamental lack of understanding about what intelligence is.

On that topic, see the original post.

But slightly on a tangent: while we're having this conversation, we need to avoid the false
1/

This is my favorite!

Eugene.

Ugh I hate those

I wonder if Cobol is in line with what you're looking for?

Admittedly, you have decades more experience than I do with this, so while I'm interested in your query, I don't have much by way of suggestion.

On the off chance you're into audio synthesis, I've heard SuperCollider is solid.

One thing to check if you haven't: if a source file has levels defined for, say, 1= "not at all satisfied" thru 5= "extremely satisfied," and nobody chose a 2, does your pkg retain the unused level on import?

I've seen sjlabelled drop an unused label, which recoded all the data down a level. 😬

Oh this is very promising.

I've used {haven} and {sjlabelled} in the past, but as good as they are, I see why my older colleagues stick with SPSS when multiple types of missings are involved. They both feel a bit like workarounds, not native features.

Yep, def a big swing, and either a triple or a home run depending on how you feel about their storytelling choices. Be prepared to feel things and you will not be disappointed 😅

I'll have to try it out!

Do it!

I think it hits an "almost perfect" game -- I don't know that I agree with either of the endings, but the game is meaningful, so surprising, and fun to play, and I respect the heck out of the developers for telling the story they did