Add metadata filter functions about posteriordb HOT 10 CLOSED

Have you a status here @paul-buerkner ?

from posteriordb.

Oh, for some reasons, I overlooked I was assigned for it. Sorry, I didn't take any further look at this.
Do you have a proposal for how to do it already? Otherwise, I will think about it a little bit more today.

from posteriordb.

Not really. I think this defines a little how we would access/filter in the databases so maybe have ashort discussion on it. I could also do the work, but the API need some thinking.

from posteriordb.

I agree. Let's think of this a little bit more and then have a discussion to make some decisions about this, maybe thursday or friday?

from posteriordb.

from posteriordb.

Sounds good.

from posteriordb.

Summary of discussion:
We implement the filter function based on the dplyr filter function, essentially using the same framework. The API will be:

pos <- pdb_filter_posterior(.pdb, ...)
ms <- pdb_filter_models(.pdb, ...)
ds <- pdb_filter_data(.pdb, ...)

Here ... is filtering syntax according to the dplyr filter function:
https://dplyr.tidyverse.org/reference/filter.html

This would mean that we would implement this as three steps:

1. create a pdb_posteriors_df() function to create a posterior/model/data tibble for all elements in the posterior database. Here we expand all keyterms.
2. Use dplyr::filter() to select from the tibble.
3. Return posteriors as read posterior objects.

from posteriordb.

Lovely, I am very happy with this interface. One minor comment. You use posterior in singular and models in plural. Does your plan to have singular/plural aliases still apply or shall we settle for just one of them?

from posteriordb.

Good point. I think using both would be reasonable (using alias).

from posteriordb.

I'm not familiar with dplyr syntax, can you show me how these queries should be written?

1. Find all posteriors where the model was added by Måns.

   My attempt

   ms <- pdb_filter_models(.pdb, added_by == "Mans Magnusson")
   posteriors <- ??

2. Find all posteriors where the model has keyword bda3_example.

3. Same as 2. but only find posteriors that have a gold standard.

from posteriordb.