rmcelreath / rethinking

Statistical Rethinking course and book package
2.16k stars 604 forks source link

Precis reports PI, not HPDI #297

Open lisamr opened 3 years ago

lisamr commented 3 years ago

The documentation of precis says that the HPDI is reported, but the PI is actually reported. See below:

library(rethinking)
x = plogis(rnorm(10000, 1, 1)) 
PI(x)
HPDI(x)
precis(x) #matchs PI, not HPDI

Would it be possible to either update the function so that HPDI is used or allow us to choose how the the intervals are summarized? Thanks!

eveskew commented 8 months ago

Just seconding this. It appears PIs are output from precis even for objects such as CmdStanFit where I'm guessing users are expecting HPDIs instead.