How to edit pp_check plot of brms?

Modeling
1

Couldn’t figure out how to edit pp_check plots.

How can I change two things on the plot?

  1. Point size of yrep on the plot?
  2. Column/bar width?

Data

library(tidyverse)
n = 100
a = tibble(sex = rep(c("m", "m", "m", "m", "m", "m", "m", "f", "f", "f"), length.out = n), year = rep(c(3, 4, 5), length.out = n))
b = tibble(sex = rep(c("m", "m", "m", "m", "m", "m", "f", "f", "f", "f"), length.out = n), year = rep(c(2, 3, 4), length.out = n))
c = tibble(sex = rep(c("m", "m", "m", "m", "m", "f", "f", "f", "f", "f"), length.out = n), year = rep(c(1, 2, 3), length.out = n))
d = tibble(sex = rep(c("m", "m", "m", "m", "f", "f", "f", "f", "f", "f"), length.out = n), year = rep(c(0, 1, 2), length.out = n))
e = rbind(a, b)
f = rbind(e, c)
df = rbind(f, d)
df = df %>% mutate(sex = as.factor(sex))
df %>% ggplot(aes(year, fill = sex)) + geom_bar(position = "fill") + ylab("proportion") + scale_fill_manual(values = c("red", "skyblue"))

Model and pp_check

m = brm(sex ~ year, family = "bernoulli", data = df)

m %>% pp_check(type = c("bars"))

image
image700×432 2.92 KB

1 Like
2

The pp_check() function in the brms package is just a ggplot object, so you can do anything that you would normally do in ggplot:

pp_check(m) +
   geom_point(size = 2)

I don’t actually know that the geom_point option there will do what you want it to, but that’s at least the general idea of how to edit the posterior predictive check plot

2 Likes
3

Beyond manipulating the resulting ggplot object, as @wgoette correctly suggests, note that pp_check delegates to the ppc_XXX functions in bayesplot which have a bunch of ways to get customized. In your case, the size and fatten parameters should be helpful: PPCs for discrete outcomes — PPC-discrete • bayesplot

Best of luck with your visuals!

2 Likes