Global Aesthetics Mappings and Side Geometries

library(ggside)
#> Loading required package: ggplot2
#> Registered S3 method overwritten by 'ggside':
#>   method from   
#>   +.gg   ggplot2

p <- ggplot(mpg, aes(displ, hwy, colour = class)) +
   geom_point(size = 2) +
   theme_bw()

The purpose of this vignette is to document a slight change in API expectation with certain geom_*side* when there are global aesthetic mappings for both x and y.

In ggside (<0.1.0), the user was expected to explicitly map all computed aesthetics, especially in the case when x and y are mapped globally for the main panel. The nature of ggside is generally to plot supplemental data, and the user will likely want to inherit the x global mapping if they use a geom_xside* geometry, but map something different to y.

This decision was made in order to keep all geom_*side* geometries as similar to their ggplot2::geom_* geometry counterparts. However, using ggside becomes a tad bit annoying when you must specify every positional aesthetic, especially when it obvious what the user may want.

In ggside (>=0.1.0), certain geom_*side* geometries will use the default aesthetic mapping of the associated Stat, depending on the orientation parameter.

For example, geom_xsidedensity will ignore the y global mapping and assign y = after_stat(density) to that layer’s aesthetic mappings if no y mapping was specified. The same can be said for geom_ysidedensity, it will ignore the x global mapping and assign x = after_stat(density) if no x mapping was specified.

This should be much less typing.

p + 
  geom_xsidedensity() +
  geom_ysidedensity()

Of course, if you would like to use a different computed variables, you will need to specify it like normal.

p + 
  geom_xsidedensity(aes(y = after_stat(count))) +
  geom_ysidedensity(aes(x = after_stat(scaled)))

You can even use geom_xsidedensity(orientation = "y") to force it to inherit the y global mapping and map x = after_stat(density). I don’t expect many users will do this, but the option is available nonetheless.

#
p + 
  geom_xsidedensity(orientation = "y") #just use geom_ysidedensity()