ggplot2 offers the fantastic option for displaying complex data in
forms of ‘many multiple’, i.e. the facets. From the example of
facet_grid:
In the above example, the panel’s borders are drawn with the default
settings of theme_bw().
If we desire the the axis lines, such as those given in this package, the distinction of the panels disappears.
library(lemon)
p <- p + coord_capped_cart(bottom='both', left='both') +
theme_bw() + theme(panel.border=element_blank(), axis.line=element_line())
p + facet_grid(drv ~ cyl)The above example is re-created below with both left- and bottom-axis lines repeated.
library(lemon)
p + facet_rep_grid(drv ~ cyl) + coord_capped_cart(bottom='both', left='both') +
theme_bw() + theme(panel.border=element_blank(), axis.line=element_line())## Coordinate system already present. Adding new coordinate system, which will
## replace the existing one.
facet_rep_grid of the lemon
package.In the following example, we change the facet from a grid to being
wrapped on the interaction of drv and cyl,
and add free scaling on y-axis. facet_wrap would
normally print the y-axis tick labels for each panel, but still ignores
the x-axis.
facet_wrap keeps y-axis label ticks
with scales='free_y'.A work around by keeping both axes free scale, and fixing the x-axis
with either scale_x_continuous or limits in
cord_*, but the same x-axis tick labels are repeated. And
this is a bit tedious.
p + facet_wrap(~ interaction(cyl, drv), scales='free') +
coord_capped_cart(bottom='both', left='both', xlim=c(2,7))## Coordinate system already present. Adding new coordinate system, which will
## replace the existing one.
We can specify which labels to keep with facet_rep_wrap.
Default is repeat.tick.labels=FALSE when
scales='fixed' which removes tick labels on all axes (shown
in earlier figure). When using free scales on
facet_rep_wrap, the appropiate labels are drawn.
repeat.tick.labels we are free
to specify which sides to keep.The argument repeat.tick.labels accepts a range of
arguments: top,right,bottom, and
left, any combination of these are accepted, as well as
TRUE and FALSE which are interpreted as all or
none, and all and none.
There are many posibilities. Examples given below (but not executed), and they might not be pretty.
p + facet_rep_wrap(~ interaction(cyl, drv), scales='free_y', repeat.tick.labels = 'all')
p + facet_rep_wrap(~ interaction(cyl, drv), scales='free_y', repeat.tick.labels = 'bottom')
p + facet_rep_wrap(~ interaction(cyl, drv), scales='free_y', repeat.tick.labels = 'left')
p + scale_x_continuous(sec.axis = dup_axis()) +
facet_rep_wrap(~ interaction(cyl, drv), scales='free_y', repeat.tick.labels = 'all')
p + scale_x_continuous(sec.axis = dup_axis()) +
facet_rep_wrap(~ interaction(cyl, drv), scales='free_y', repeat.tick.labels = c('top','left'))