Hi -
I am using a modified version of the LDA model in the Stan reference model (changed word distribution, added document-level covariates). Is there an established procedure for adding models to the manual? I couldn’t find one in my Google searches.
Thanks much,
Bob
the manuals have been moved to repo: https://github.com/stan-dev/docs
you’re welcome to file an issue to add your model to the manual. if you’re comfortable using R markdown and bookdown (requires knitr, pandoc, minimum, and latex to generate pdfs), then you can create a branch and submit a PR. otherwise, write it up in the issue and we’ll try to get it in there.
Great, thanks!
I think the user’s guide will is the right place for a new model if it illustrates a general point. Otherwise, I think a case study might be more appropriate.
Having said that, if you submit a PR for the user’s guide, we’ll probably take it. We’d like to encourage more contributions!
I’d like to see this example! Has it been created yet?
I’m not sure what you’re asking for—there are a lot of example models in the user’s guide and lots of user’s guide pull requests.
I was interested in the LDA model with covariates on documents and alternative distributions on the words as mentioned by user saudiwin. The current LDA examples are great, but I was interested in seeing this extension.
Just following up if you managed to document this modified version of the LDA model?
Hi Skipton -
Yes I have working Stan code if you’d like to play around with it. I won’t be sending out any case studies or manual additions for a while as we are still probing the model and finding the best way to do constraints/identification. But I’m happy to share what we have now if you email me at rmk7@nyu.edu
Best,
Bob