mvuorre / brmstools Goto Github PK

It would be useful to be able to transform the parameters draws before plotting, e.g. if one have estimated a simple logistic model with a random intercept (with no other terms), it would be useful to get estimates on the probability scale. (inverse logit)

The plotting functions of this package rely heavily on two poorly coded functions; tidycoef() and tidyfitted(). Consequently the code is unnecessarily complicated and slow.

The code could be made significantly more straightforward and tidy by relying on functions in the wonderful tidybayes package instead.

Hi Matti, love the package, really useful for plotting quickly. I'm wondering if you plan to offer support for plotting the random effects from multi-level ordinal models computed using brms? The following error pops up when I attempt to implement directly:

Error in coef(model, summary = FALSE)[[grouping]][, , parameter] : 
  subscript out of bounds

I'm assuming by the error that this is due to the nature of how the ordinal effects are stored in brms? (Running R 3.5.0 in RStudio and just reinstalled the dev version of brmstools today)

Let say I have a regression model like this: Y ~ (1|A/B), then I can't specify the interaction (A:B) in the grouping argument of the forest function without error

Error: object 'A' not found

with the call:

forest(model, grouping = "A:B")

Hello,

Thanks for the awesome package! However, I'm getting an error when I try to use the spaghetti plot and panels plot functions when I fit a model with a random intercept and slope. Wondering what I might be missing?

library(brmstools)

fit1 <- brm(count ~ log_Base4_c + Trt + (log_Age_c|patient), 
            data = epilepsy, family = poisson())

spaghetti(fit1, xvar="log_Base4_c")
>Error in eval(predvars, data, env) : object 'log_Age_c' not found

panels(fit1, xvar="log_Base4_c")
>Error in eval(predvars, data, env) : object 'log_Age_c' not found

Thanks for any insight you can provide!

I am trying the following code, but getting a weird error message, what I am doing wrong?
I am getting this error message: Error in FUN(X[[i]], ...) : object '2.5%ile' not found

simpelmodel1<-brm(popular~1 + extrav+(1+extrav|class), data=popular2data, warmup = 1000, iter =5000, chains = 2, inits= "random", control = list(adapt_delta = 0.96),save_all_pars = TRUE, cores= 2) 

spaghetti(simpelmodel1, xvar = "extrav")

This is thus a super simple model for demonstration purposes.
The forest function does work.

Hi Matti,

Thanks for the package. I noticed you do not support visualizations with multiple grouping factors.
The phylogenetic meta-analysis model comes to mind: https://cran.r-project.org/web/packages/brms/vignettes/brms_phylogenetics.html

Manuel