Specifying a positive distribution for prior coefficients

Interfaces rstanarm
1

Hi, I’m new to Rstanarm and I’m trying to model a regression where the priors have a positive distribution like a Gamma or Exponential.

Reading through the priors Documentation, I see that there is an option for an exponential prior but when I try specifying it I get the error “The prior distribution should be one of normal, student_t, cauchy, hs, hs_plus, laplace, lasso, product_normal”.

Is there another way to do this?

On the sample dataset mpg, I wish to model the coefficients with positive priors:

post4 <- 
  stan_glm(
    hwy ~ displ + year + cyl,
    data =  mpg,
   prior = exponential(rate = 1),
    seed = 12345)
post4
prior_summary(post4)
  • Operating System:
    “Windows” “10 x64” “build 19044”
  • rstanarm Version:
    ‘2.21.3’

Thanks!