Felipe Fontana Vieira

Literally this morning, my gf: If you hadn't breathed so loud, I wouldn't have had a nightmare.

Fair 😆. The ones I just looked up:

- Maximum likelihood (ML) estimation (..) part 1, 2, and 3
- ML - Cramer Rao Lower Bound Intuition
- Least squares as ML estimator (two videos)

Matrix calculus sucks

Ben Lambert has some short videos!

I am increasingly getting comfortable with the idea that estimation and uncertainty quantification can be treated in quite different ways, which would once have been quite strange to me.

I'd love to read an essay by you on this topic.

When I took a similar path, I realized the toolkits were actually chasing different abstract quantities. "Probability," it turns out, has myriad meanings, even when it refers to the same data.

Tbh, it took me a few courses and around a year to realize that a lot of it boils down to regularization/optimization. There is always the "philosophical" (broadly speaking) avenue to it, which is were I guess most "heated" stuff comes up

Does anyone know an example (paper, blog post) where frequentist and Bayesian methods lead to different conclusions?

#rstats #stats

Or an anecdote that lead you to embrace one or the other approach?

Just received a mysterious e-mail inquiring whether the silkie bantam rooster is still available and to my disappointment, this apparently is not some sort of code word.
Looks like there's another Julia Rohrer in a different part of Germany who is selling chicken.

This 21st century Fontana is confident after creating fake data already

Much of expertise in psych. comes from supposedly “data-supported” conclusions. Yet a lot of it is quite sloppy and concerning. Plus, we generally agree that statisticians shouldn’t make claims about emotions, but suddenly we accept psychologists claiming big things achieved through statistics

Tbh, I should have said “research-oriented psychologists“ (still something very general though). I also definitely believe in collaboration. However, the problem I generally see and tried to target with my statement:

But I get your point!

In psychology, we have “normalised“ too much that we “teach students to be psychologists, not statisticians“.

There are many layers to this statement. I do see your point, despite disagreeing with it. What I am wondering though: one can teach statistics through SAS/JASP/SPSS, so what exactly draws you to something like R?

I will never get used to the disdain our country holds for brown and Black children.

From Palestine to Iran to Sudan, we normalize the destruction of schools and hospitals and daycares. We reduce children to death tolls and their futures to rubble with no consequence or second thought.

And I really hope people eventually get to it. rix gets you so far with such a little effort, but it will probably take some years (even among the people who supposedly care about *full* reproducibility)

At least for the packages, you can use remotes::install_version and write a function around it with regex that automatically downloads everything. But it is one of those things that I would be very surprised if it worked smoothly

You mean a function that would automatically install the packages from the sessionInfo output?

That is what we call being baptised on fire after studying in Leuven

This is Belgium and one can imagine you trying to explain to US professors during PhD applications that a 16 here is good, while the other applicant might have a 4.1 GPA

Evaluating Local Structural-After-Measurement (LSAM) and Traditional Approaches for the Estimation of Complex Nonlinear Effects Among Latent Variables: https://osf.io/xk8um

This is quite exciting

In T, pipelines are first-class, and it's possible to run R or Python code. Right the pipeline code, with some nodes running #RStats code, and other #Python code.

oh no, here we go again

There's a nice website that walks through some of the choices: choosealicense.com

I don't understand how someone can be confident with statistics after consultancy work. This week happened again: a question, you think it will take 1 hour, but it takes 1 day (and counting), and you find a bug in the R package. All of that about a topic and package that I supposedly know a lot.

CC by 4.0 is fine!