Replicate the horseshoe prior implementation of Bayesian SCM

marcdotson / category-captainship

An empirical generalization of the category captainship paper.

MIT License

1 stars 1 forks source link

Replicate the horseshoe prior implementation of Bayesian SCM #78

Closed marcdotson closed 2 years ago

marcdotson commented 2 years ago

@morganbale @cdbale note that as part of the cleanup I've moved Model into Code. I know that Code is very much in need of a cleaning up at some point. If you can, just move the code you're working on into the root of the Code directory with any of the executables and other source files into a folder called Source within Code.

marcdotson commented 2 years ago

@morganbale good job cleaning up the files. I've pulled #81 after cleaning up a few merge conflicts in some of the Stan files. Hopefully I didn't mess anything up? See the closed PR for details on what I changed.

I've also modified the main branch by moving your files into Writing and removing the figures from Code. I went to knit your files and discovered that you named all the code chunks the same. That's like naming all of the files the same -- you'll need a different name for each code chunk for the file to knit and for the figures to have unique names.

I'd recommend creating a new branch to knit and finalize where we're at in terms of the three models: bscm_intercept, bscm_level1, and bscm_testing.

marcdotson commented 2 years ago

@morganbale I can take a shot at this now. Good luck to you and Cameron finishing things up for the semester!

morganbale commented 2 years ago

@marcdotson sounds good. Thanks! Do you need me to rename the chunks or you will do that as you go through?

marcdotson commented 2 years ago

@morganbale I'll take care of it as I go through.

marcdotson commented 2 years ago

@morganbale I cleaned it up a bit in the replicate-bscm branch:

I renamed them so I can remember the order: 01_bscm-horseshoe.Rmd, 02_bscm-intercept.Rmd, and 03_bscm-levels.Rmd.
I was able to knit 01_bscm-horseshoe.Rmd and 03_bscm-levels.Rmd, but it looks like there is an issue with the model used in 02_bscm-intercept.Rmd -- bscm_b1_intercept.stan.

Will you take a look at bscm_b1_intercept.stan? If you can get it working, go ahead and knit 02_bscm-intercept.Rmd. Walking through each of these will be a good review when we meet tomorrow.

marcdotson commented 2 years ago

Merged with PR #82.