Highlight extreme points in NPDE plots

[timwaterhouse]

Since NPDEs are constrained to be within a certain range (depending on the number of simulation replicates), it would be nice to be able to highlight the points that are sitting on the edges of that range.

For k replicates, I believe the extreme situations (i.e., simulated values are either all above or all below observed) are handled by setting NPDE to qnorm(1/(2*k)) or qnorm(1 - 1/(2*K)). It could be helpful to identify which points correspond to those values in the plots. I'm thinking the output could look something like this:

library(pmplots)
#> Loading required package: ggplot2
df <- pmplots_data_obs()
p <- npde_time(df)
k <- 150
df_extreme <- dplyr::filter(
  df,
  abs(round(NPDE, 3)) == round(qnorm(1 - 1/(2*k)), 3)
)
p + geom_point(data = df_extreme, fill = "red", pch = 23, cex = 2)
#> `geom_smooth()` using formula 'y ~ x'

^{Created on 2021-06-28 by the reprex package (v2.0.0)}

Unfortunately it would mean having to supply the value of k, so the user would need to be careful about that.