stat_summary_hex not working properly with discrete z values

[rvalieris]

I found a problem with stat_summary_hex when z is a discrete variable:

p1 = ggplot(diamonds, aes(x=carat, y=depth)) + geom_hex()
p1 # ok, default stat

p2 = ggplot(diamonds, aes(x=carat, y=depth)) + stat_summary_hex(aes(z=carat),fun=length)
p2 # ok, same result as p1, carat is a continuous variable 

p3 = ggplot(diamonds, aes(x=carat, y=depth)) + stat_summary_hex(aes(z=cut),fun=length)
p3 # bug, cut is a discrete variable

I expected p3 to be the same as p1 and p2 since length should return the same number regardless of which variable I pick for z, however it seems that the bins are being grouped by z, which was not intended.

Also stat_summary_2d have the same behavior, but its harder to see the overlapping bins.

this was tested with ggplot2 3.4.2

[teunbrand]

I think this is one of those cases similar to the ones documented in ?aes_group_order where the defaults fail. In any case, it is true in (almost?) all geoms that using a discrete variable auto-implies grouping and I don't think it would make sense to apply a specific exception here. The plot can be fixed by using aes(group = 1) in either the main plot call or the stat_summary_hex().

[rvalieris]

Thanks, that The 'group' aesthetic is by default set to the interaction of all discrete variables in the plot. was missing for me, I did tried to set group=NULL but didn't see any difference, group=1 solves the issue.