From 96902a7eaba0106169bfa7bf4c5b461932ef3440 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Paul-Christian=20B=C3=BCrkner?= Date: Thu, 12 Sep 2024 13:14:00 +0200 Subject: [PATCH] fix issue #1655 --- R/priors.R | 8 ++++++++ man/R2D2.Rd | 6 +++++- man/horseshoe.Rd | 4 ++++ 3 files changed, 17 insertions(+), 1 deletion(-) diff --git a/R/priors.R b/R/priors.R index d6acd26a1..83d7fad62 100644 --- a/R/priors.R +++ b/R/priors.R @@ -2051,6 +2051,10 @@ eval_dirichlet <- function(prior, len = NULL, env = NULL) { #' See the documentation of \code{\link{brm}} for instructions #' on how to increase \code{adapt_delta}. #' +#' The prior does not account for scale differences of the terms it is +#' applied on. Accordingly, please make sure that all these terms have a +#' comparable scale to ensure that shrinkage is applied properly. +#' #' Currently, the following classes support the horseshoe prior: \code{b} #' (overall regression coefficients), \code{sds} (SDs of smoothing splines), #' \code{sdgp} (SDs of Gaussian processes), \code{ar} (autoregressive @@ -2148,6 +2152,10 @@ horseshoe <- function(df = 1, scale_global = 1, df_global = 1, #' See the Examples section below. #' #' @details +#' The prior does not account for scale differences of the terms it is +#' applied on. Accordingly, please make sure that all these terms have a +#' comparable scale to ensure that shrinkage is applied properly. +#' #' Currently, the following classes support the R2D2(M2) prior: \code{b} #' (overall regression coefficients), \code{sds} (SDs of smoothing splines), #' \code{sdgp} (SDs of Gaussian processes), \code{ar} (autoregressive diff --git a/man/R2D2.Rd b/man/R2D2.Rd index 408f7c393..5f03c9147 100644 --- a/man/R2D2.Rd +++ b/man/R2D2.Rd @@ -31,7 +31,11 @@ Function used to set up R2D2(M2) priors in \pkg{brms}. The function does not evaluate its arguments -- it exists purely to help set up the model. } \details{ -Currently, the following classes support the R2D2(M2) prior: \code{b} +The prior does not account for scale differences of the terms it is + applied on. Accordingly, please make sure that all these terms have a + comparable scale to ensure that shrinkage is applied properly. + + Currently, the following classes support the R2D2(M2) prior: \code{b} (overall regression coefficients), \code{sds} (SDs of smoothing splines), \code{sdgp} (SDs of Gaussian processes), \code{ar} (autoregressive coefficients), \code{ma} (moving average coefficients), \code{sderr} (SD of diff --git a/man/horseshoe.Rd b/man/horseshoe.Rd index 035557ec1..5e9c83138 100644 --- a/man/horseshoe.Rd +++ b/man/horseshoe.Rd @@ -103,6 +103,10 @@ The horseshoe prior is a special shrinkage prior initially proposed by See the documentation of \code{\link{brm}} for instructions on how to increase \code{adapt_delta}. + The prior does not account for scale differences of the terms it is + applied on. Accordingly, please make sure that all these terms have a + comparable scale to ensure that shrinkage is applied properly. + Currently, the following classes support the horseshoe prior: \code{b} (overall regression coefficients), \code{sds} (SDs of smoothing splines), \code{sdgp} (SDs of Gaussian processes), \code{ar} (autoregressive