Closed cgrandin closed 10 months ago
I use those all the time when I need to 'hand calculate' a posterior median for a metric that we don't automatically do it. I'm fine to get rid of them as csv's as long as the information is stored somewhere.
What if we put them as data frames attached to the RDS file? Then you would access like this:
base_model <- readRDS("/path/to/01-base.rds") base_model$keyposteriors base_model$nuisanceposteriors
If you had built the doc you wouldn't even need to run the loading part, they would just be there.
They aren't used anywhere and just seem to pop up in bugs all the time. Please speak up if you have a reason to keep these!!