to CRAN

DESCRIPTION

@@ -24,7 +24,8 @@ Suggests:
     testthat
 Depends:
     bayesplot,
-    R (>= 3.4.0)
+    R (>= 3.4.0),
+    rstan (>= 2.26.0)
 Imports:
     coda,
     dplyr,
@@ -36,7 +37,6 @@ Imports:
     Rcpp (>= 0.12.9),
     RcppParallel,
     rlang,
-    rstan (>= 2.26.0),
     rstantools (>= 2.0.0)
 LinkingTo:
     BH (>= 1.66.0),

NAMESPACE

@@ -34,6 +34,7 @@ importFrom(RcppParallel,CxxFlags)
 importFrom(RcppParallel,RcppParallelLibs)
 importFrom(bayesplot,color_scheme_get)
 importFrom(bayesplot,pp_check)
+importFrom(bayesplot,ppc_ribbon)
 importFrom(bayesplot,theme_default)
 importFrom(coda,spectrum0.ar)
 importFrom(dplyr,group_by)

R/pp_check.R

@@ -6,18 +6,18 @@
 #' For other types of posterior predictive checks for example with `bayesplot`, 
 #' you can extract the variable `yrep` from the output, see examples.
 #' 
-#' @importFrom bayesplot pp_check
+#' @importFrom bayesplot pp_check ppc_ribbon
 #' @param object An output from [walker()].
 #' @param ... Further parameters to [bayesplot::ppc_ribbon()].
 #' @export
 #' @examples 
 #' \dontrun{
 #' # Extracting the yrep variable for general use:
 #' # extract yrep
-#' y_rep <- rstan::extract(object$stanfit, pars = "y_rep", permuted = TRUE)$y_rep
+#' y_rep <- extract(object$stanfit, pars = "y_rep", permuted = TRUE)$y_rep
 #' 
 #' # For non-gaussian model:
-#' weights <- rstan::extract(object$stanfit, 
+#' weights <- extract(object$stanfit, 
 #' pars = "weights", permuted = TRUE)$weights
 #' y_rep <- y_rep[sample(1:nrow(y_rep), 
 #'   size = nrow(y_rep), replace = TRUE, prob = weights), , drop = FALSE]

vignettes/walker.Rmd

@@ -72,14 +72,13 @@ The output of `walker` is `walker_fit` object, which is essentially a list with
 
 ```{r pars}
 print(fit$stanfit, pars = c("sigma_y", "sigma_rw1"))
-library(bayesplot)
 mcmc_areas(as.matrix(fit$stanfit), regex_pars = c("sigma_y", "sigma_rw1"))
 ```
 
 Let's check how well our estimates of $\beta$ coincide with the true values (the solid lines correspond to true values):
 
 ```{r plot_with_true_betas}
-betas <- summary(fit$stanfit, "beta_rw")$summary[, "mean"]
+betas <- rstan::summary(fit$stanfit, "beta_rw")$summary[, "mean"]
 
 ts.plot(cbind(rw, beta1, beta2, matrix(betas, ncol = 3)),
   col = rep(1:3, 2), lty = rep(1:2, each = 3))
@@ -138,7 +137,7 @@ $$
 This is essentially local linear trend model [@harvey] with restriction that there is no noise on the $\beta$ level. This model can be estimated by switching `rw1` function inside of the walker formula to `rw2`, with almost identical interface, but now $\sigma$ correspond to the standard deviations of the slope terms $\nu$. The Gaussian prior for $\nu_1$ most also be defined. Using RW2 model, the coefficient estimates of our example model are clearly smoother:
 
 ```{r walker_rw2}
-fit_rw2 <-walker(y ~ -1 + 
+fit_rw2 <- walker(y ~ -1 + 
     rw2(~ x1 + x2, beta = c(0, 10), sigma = c(2, 0.001), nu = c(0, 10)), 
   refresh = 0, init = 0, chains = 1, sigma_y = c(2, 0.001))
 plot_coefs(fit_rw2, scales = "free") + ggplot2::theme_bw()