Error when projecting with non-NULL solution_terms.

[fweber144]

Applying project() with non-NULL argument solution_terms leads to an error. I'll create a PR for fixing this. Below is a reproducible example.

# Based on <https://github.com/stan-dev/projpred/blob/master/README.md>.

library(projpred)
library(rstanarm)
options(mc.cores = parallel::detectCores(logical = FALSE))

data("df_gaussian", package = "projpred")
df_gaussian <- df_gaussian[1:40, ]
mydat <- cbind("y" = df_gaussian$y, as.data.frame(df_gaussian$x))
mydat$mygroup <- gl(n = 8, k = floor(nrow(mydat) / 8), labels = paste0("gr", seq_len(8)))
set.seed(457211)
mydat$noise <- rnorm(nrow(mydat))
mygroup_icpts_truth <- rnorm(nlevels(mydat$mygroup), sd = 6)
mygroup_V1_truth <- rnorm(nlevels(mydat$mygroup), sd = 6)
mydat$y <- mydat$y + mygroup_icpts_truth[as.numeric(mydat$mygroup)]
mydat$y <- mydat$y + mygroup_V1_truth[as.numeric(mydat$mygroup)] * mydat$V1

myfit <- stan_glmer(y ~ V1 + V2 + V3 + V4 + V5 + noise + (1 + V1 | mygroup),
                    data = mydat,
                    seed = 1140350788)

cvs <- cv_varsel(myfit)
solution_terms(cvs)

C_proj <- project(myfit,
                  solution_terms = solution_terms(cvs)[1:3],
                  ndraws = ceiling(0.05 * nrow(as.matrix(myfit))))

The last call throws an error:

Error in project(myfit, solution_terms = solution_terms(cvs)[1:3], ndraws = ceiling(0.05 *  :
  solution_terms contains an index larger than the number ofvariables in the model.