Setting the last plot in the middle with facet_wrap

Question

I am trying to create some multiplots with facet_wrap. However I am not sure if is the right approach for my graph. Here is a short reproducible example:

 ggplot(airquality, aes(x = Day, y = Temp)) +
  facet_wrap(~Month) + 
  geom_line()

This produces this plot here:

Is it possible to "center" the two plots in the 2nd row with the facet_wrap approach ?

Note, that I don't want to reorder the plots, I just want to Center the 2nd row

Liquefy

You're Right. I added a short example. Note, that I dont want to reorder the plots. I want to center them – Liquefy

I have un-downvoted. now unfortunately both @Jaap and I had misunderstood your question and marked it as a duplicate. Please ask a new question, because otherwise no one will pay attention to this question. I wonder, however, what you would like to achieve with 'centering'. I don't think that this is in the intention of the ggplot developers, and also contradicts a bit the 'facet style' - you would need the labelled x-axis on both facet rows. Therefore, probably best to use one of the plot combining packages such as cowplot. Or simply use facet_grid – Commonality

The question is open again. – Agamete

Answer 1

@Tjebo's suggestion of using cowplot will work:

p <- ggplot(mapping = aes(x = Day, y = Temp)) +
  facet_wrap(~Month) + 
  geom_line()

cowplot::plot_grid(
  p %+% subset(airquality, Month < 8),
  p %+% subset(airquality, Month > 7),
  nrow = 2
)

Agamete

Didn't know that this would work p %+% subset(airquality, Month < 8). (+1) – Cadmium

The cowplot solution is very nice ! Just one question. Is there way to assign one jont axis title. In particular to have one "Temp" in the middle instead of two – Liquefy

?cowplot::draw_label – Agamete

Answer 2

You might also consider the function set_panel_size from the egg package which lets you set the panel size (width, height) of multiple plots to an absolute measurement, see this vignette for more details.

Using Axeman's code to create the plots

library(egg)
library(gridExtra)
p <- ggplot(mapping = aes(x = Day, y = Temp)) +
  facet_wrap(~Month) + 
  geom_line()

p1 <- p %+% subset(airquality, Month < 8) + labs(x = NULL)
p2 <- p %+% subset(airquality, Month > 7)

Now arrange the plots using grid.arrange after specifying the panel sizes

grid.arrange(grobs = lapply(
  list(p1, p2),
  set_panel_size,
  width = unit(5, "cm"),
  height = unit(4, "cm")
))

Cadmium

Yes, this might very well be the better option, depending on the use case. – Agamete

Answer 3

And yet another option to achieve the desired result would the ggh4x package which via ggh4x::facet_manual adds some flexibility to position the facets via the design argument:

library(ggplot2)
library(ggh4x)

base <- ggplot(airquality, aes(x = Day, y = Temp)) +
  geom_line()

design <- c(
"
AABBCC
#DDEE#
"
)

base + ggh4x::facet_manual(~Month, design = design)

design <- c(
"
AABBCC
DDDEEE
"
)
base + ggh4x::facet_manual(~Month, design = design)

Wendelin

Answer 4

And just in case anyone wants to know how to do it from scratch in grid...

library(ggplot2)
library(dplyr)
library(grid)

vps <- c("top_left", "top_mid", "top_right", "bottom_left", "bottom_right")

main_vp  <- vpTree(viewport(name = "main"), vpList(
              viewport(x = 1/6, y = 0.75, width = 1/3, height = 0.5, name = vps[1]),
              viewport(x = 3/6, y = 0.75, width = 1/3, height = 0.5, name = vps[2]),
              viewport(x = 5/6, y = 0.75, width = 1/3, height = 0.5, name = vps[3]),
              viewport(x = 1/3, y = 0.25, width = 1/3, height = 0.5, name = vps[4]),
              viewport(x = 2/3, y = 0.25, width = 1/3, height = 0.5, name = vps[5])))

grid.newpage()
pushViewport(main_vp)

plots <- lapply(1:5, function(i){ 
  seekViewport(vps[i])
  invisible(grid.draw(ggplotGrob(
    
    ggplot(filter(airquality, Month == levels(as.factor(airquality$Month))[i]), 
           mapping = aes(Day, Temp)) + 
      geom_line() + 
      facet_grid(.~Month)
    
)))})

^{Created on 2020-07-03 by the reprex package (v0.3.0)}

Xymenes