Sean Pinkney retweeted
Dear colleagues, I'm in the last year of my postdoc at @FlatironInst and I'm looking for research and faculty positions in statistics, data science, and machine learning (for the summer/fall of '25).
DM me if you know of places which would be a good fit!
Sean Pinkney retweeted
The second post in my series on copulas in @mcmc_stan is up.
This one defines the Gaussian copula and shows how to use it in a Stan model with a small example.
bggj.is/posts/gaussian-copul…
In Listening to Prozac, Kramer writes about how some depressed people respond to Prozac in a hypomanic way, but it doesn't seem to cause them any issues. They have loads of energy, getting things done, grabbing life by the balls, etc. Start there and find out why.
what if we created new, better anti-depressants for healthy people which gave you _negative_ levels of depression?
- instead of "not suicidal" go for "intense lust for life"
- instead of "gets out of bed" go for "endless energy"
- instead of "rarely sad" go for "always happy"
253
@mcmc_stan thank you for a wonderful Stancon! A special thanks to the organizers (@charlesm993, @liza_p_semenova, @michael_c0llyer, @EttieUnwin, @josswright, @rlmcelreath, Ben Lambert, @flaxter, @willpearse, and @MakkundaSharma)! Everyone who attended (links to slides next) 1/2
1
14
834
and all the wonderful people I met and all the wonderful people I've gotten to see again!
Talk slides: github.com/spinkney/open-tal…
Tutorial slides: github.com/spinkney/hierarch…
2/2
Oh shit the Bayesians have arrived in the UK, my priors are recalibrated and ready for some boat disasters
1
9
992
Join the tutorial to also get some fun graphs and the code to make them
StanCon24 Tutorial!
Get excited for
a prior for partially centered parameters (i.e. 0 < w < 1), loc-scale meta-analysis on bayes vs frequentist, a data driven prior for marginalizing out id-level hierarchy when you want Bayes to go vroom.
@mcmc_stan
1
4
1,270
StanCon tutorials thread discourse.mc-stan.org/t/stan…
StanCon24 Tutorial!
Get excited for
a prior for partially centered parameters (i.e. 0 < w < 1), loc-scale meta-analysis on bayes vs frequentist, a data driven prior for marginalizing out id-level hierarchy when you want Bayes to go vroom.
@mcmc_stan
Github repo at github.com/spinkney/hierarch…
All code and slides will be there before the tutorial Friday, Sept. 13
I'm making the hierarchical models tutorial at @mcmc_stan-con and it's coming along great in quarto
I want to apologize to all the stuff I'm supposed to be working on, I did the hypoexponential thing in the car on the way to camping. Phone is about to die. See you all Monday!
1
3
341
Coding the hypoexponential lpdf in a stable way in @mcmc_stan . Signed versions of logsumexp and other fun stuff.
See discourse.mc-stan.org/t/mini…
Many PDF calculations I found created a huge bidiagonal matrix for this which is quite inefficient. I handle it with vectors all on the log space.
1
149