Meeting chat log
00:11:21 Daniel Chen: is this like picking the parameters for a given distribution? like beta(0, 1) vs beta(0.1, 0.9)?
00:12:09 Daniel Chen: ok nvm this is the svm example right now. I guess that question is for the Bayesian point
00:12:49 Jiwan Heo: I think so, updating priors to make posterior distribution
00:31:36 Daniel Chen: the usemodels is mainly a way to help you out with somewhat reasonable defaults
00:35:02 Daniel Chen: like the "caterpillar plot" to help you see if things converged?
00:35:08 Jiwan Heo: yea!
00:38:13 Daniel Chen: i forgot what data is using? the post resamples?
00:38:23 Daniel Chen: you might need to manually pull that out from the data?
00:44:57 Daniel Chen: for the bayes stuff. is there a way to change the backend for the bayes calculation? e.g., using STAN or something?
00:47:33 Jiwan Heo: maybe in the model spec?
00:47:43 Jiwan Heo: bayes.model = linear_reg() %>%
set_engine(engine = "stan",
prior_intercept = prior.dist,
prior = prior.dist) %>%
set_mode(mode = "regression")
00:49:08 Daniel Chen: oooh. yeah. okay
00:54:52 Daniel Chen: yeah i guess this ends up being hard since bayes is anaother layer of things to understand the examples
00:55:07 Daniel Chen: i do wonder if the SVM example or something can/should be simplified to be used as a chapter 0
00:55:26 Daniel Chen: I've only really used grid search in the past personally in the past
00:55:36 Daniel Chen: but i was before tidymodels and all hand coded
00:55:42 Daniel Chen: expand.grid was a friend :)
01:03:16 Jiwan Heo: have to jump off, thank you for the presentation!
01:09:27 Daniel Chen: i'm good for next week