The problem: I have several facets of different widths, and I can't get them to display legibly on the same plot.
The data frame:
> df
# A tibble: 361 × 6
Letter X Year ID NOTID tot
<chr> <chr> <chr> <dbl> <dbl> <dbl>
1 A x1 2019 2 2 4
2 A x1 2020 1 0 1
3 B x1 2019 2 1 3
4 B x1 2020 1 7 8
5 B x1 2021 1 0 1
6 C x1 2014 1 0 1
7 C x1 2020 1 1 2
8 C x1 2021 2 0 2
9 C x1 2015 0 1 1
10 C x1 2016 0 3 3
# … with 351 more rows
The facets are the Letters (n=27),
x axis is Year (n=25 but not all facets have all years --> x can be very different width),
y axis is ID/tot
and fill is X (n=4).
I have tried different methods:
1: with facet_wrap
a)
df_plot <-na.omit(df)%>%
ggplot(aes(x=Year, y= ID/tot , fill = X))+
geom_bar(stat = "identity", position= "dodge")+
facet_wrap(~Letter,
scales= "free_x")+
geom_label(aes(y=0.1, label= paste(round(ID, 1), "/", round(tot, 1), "\n", "=", round((ID/tot)*100, 1),"%")),
position = position_dodge(0.9),
color="black",
size=2)+
scale_y_continuous(labels = scales::percent) ; df_plot
--> the facets have all the same width:
b) Then I added:
gp <- ggplotGrob(df_plot)
# optional: take a look at the grob object's layout
gtable::gtable_show_layout(gp)
# get gtable columns corresponding to the facets (5 & 9, in this case)
facet.columns <- gp$layout$l[grepl("panel", gp$layout$name)]
# get the number of unique x-axis values per facet (1 & 3, in this case)
x.var <- sapply(ggplot_build(df_plot)$layout$panel_scales_x,
function(l) length(l$range$range))
# change the relative widths of the facet columns based on
# how many unique x-axis values are in each facet
gp$widths[facet.columns] <- gp$widths[facet.columns] * x.var
# plot result
grid::grid.draw(gp)
which was a pretty good solution for each facet to have its width, but the problem is that it puts them all on one row, so it's unreadable ...
c) So I did the same thing as above but added nrwo = 4 in facet_wrap and I obtained this plot:
It's getting closer but we can see that it's not yet the right width for each facet...
2: I tried with facet_grid, but even if it seems that each facet have its width, they are all in the same row so it's unreadable...
ggplot(aes(x=Year, y= ID/tot , fill = X))+
geom_bar(stat = "identity", position= "dodge")+
facet_grid(~Letter,
scales= "free_x",
space = "free")+
geom_label(aes(y=0.1, label= paste(round(ID, 1), "/", round(tot, 1), "\n", "=", round((ID/tot)*100, 1),"%")),
position = position_dodge(0.9),
color="black",
size=2)+
scale_y_continuous(labels = scales::percent) ; df_plot
3: And I tried unsuccessfully facet_row and facet_col (ggforce package)... (same kind of result than above).
Does anyone have a solution?... :)