We have some work to develop a Stan model. The data we are considering are relatively numerous (potentially more than 300 million data) and natively stored in a postgres database, which is continually evolving such that we want to run the Stan model at regular intervals. We can, of course, translate the data into whatever format we like (eg json) and then point Stan at the data, but is there a pre-existing/recommended strategy for getting the data out of the database and directly into Stan? I’m also interested in people think this is worth worrying about.
Cheers
Simon