No warning in stat_ydensity when dropping groups

[andrewwbutler]

Hi,

I noticed the following behavior in geom_violin when there are fewer than three data points in a given violin. In this scenario, one of the violins gets dropped without a warning.

library(ggplot2)
table(mtcars$cyl, mtcars$am)
#>    
#>      0  1
#>   4  3  8
#>   6  4  3
#>   8 12  2
ggplot(mtcars, aes(x = factor(cyl), y = mpg)) + 
  geom_violin(scale = "width", aes(fill = factor(am))) 

^{Created on 2020-07-01 by the reprex package (v0.3.0)}

It seems a related issue was addressed in stat_density in #2142 . I would propose that stat_ydensity be updated similarly here

ggplot2/R/stat-ydensity.r

Line 74 in cb0cfd2

if (nrow(data) < 3) return(new_data_frame())

Happy to put together a small PR for this if people agree.

[clauswilke]

That seems reasonable.