Mcmc_plot intervals with thicker CI lines?

Interfaces brms
1

I’m plotting the posterior distribution intervals of my model but when I paste the plot in Word the credible interval lines are too thin and you can barely see them. How can I make them thicker? and the dot showing the mean/median bigger? I have customized other parts with ggplot but I can’t figure out how to change this. Thanks.

  • Operating System: Mac
  • brms Version: 2.13.0
3

Have you looked at the vignette? The probs and probs_outer arguments seem to be what you’re looking for. They should work with mcmc_plot() as well

4

yes I have, but that only takes care of the percentage for each interval. I don’t think you can change the thickness of the lines or the size of the dot for that matter.

5

Oh wow, I totally misread your question.

Looking at the function’s source code, it appears that there isn’t a way to do this. If you’re feeling adventurous, you could probably make a custom version of the function, modify the size argument of args_inner and args_outer, and put bayesplot::: in front of any function that R doesn’t recognize.

2 Likes
6

OK, I will try that! Thanks

7

I’m not opposed to adding this to bayesplot (it might just have been an oversight originally). So if you want to open an issue on github with a feature request or even submit a pull request (if you do end up making this change yourself like @Christopher-Peterson suggested) then that would definitely be welcome!

2 Likes
8

Thanks, I just submitted a feature request. How do I know when this feature is added? As I’m writing a paper and I need this plot, thank you!

9

I had a free moment now, so I just added this on a branch in the bayesplot github repo. I responded over on GitHub but I’ll put it here too for anyone else interested on the forums:

If you install that branch of bayesplot using

devtools::install_github("stan-dev/bayesplot", ref = "mcmc_intervals-size")

then you should be able to use the new arguments inner_size, outer_size, and point_size. Does that do what you were looking for? Sometime soon I’ll make this into a pull request and so we can get it into the next release, but you should be able to use the version on that branch in the meantime.

2 Likes