Remove some of the axis labels in ggplot faceted p

问题:

I have created a plot like the one here with ggplot2 package and facet_wrap function, and I would like to suppress some of the x-axis text to make it more legible.

For example, here it would be more legible if the x-axis scales appeared only on boxes D, F, H and J.

How could I do that? Thanks in advance!

EDIT : the reproducible code

library(ggplot2)
d <- ggplot(diamonds, aes(carat, price, fill = ..density..)) +
  xlim(0, 2) + stat_binhex(na.rm = TRUE) + theme(aspect.ratio = 1)
d + facet_wrap(~ color, nrow = 1)

回答1:

If you're willing to work at the grid/grob-level, it's definitely doable.

First, we assign a ggplot object with your faceted plot

my_plot <- d + facet_wrap(~ color, nrow = 1)

Then, we load up gtable so we can use/manipulate the lower-level objects.

library(gtable)

## Loading required package: grid

Now, we extract the ggplot object into a TableGrob (apologies for the long-ish output, but I think it helps show the underlying structure of the facet plots):

plot_tab <- ggplotGrob(my_plot)
print(plot_tab)

## TableGrob (8 x 25) "layout": 33 grobs
##     z         cells       name                                    grob
## 1   0 ( 1- 8, 1-25) background          rect[plot.background.rect.263]
## 2   1 ( 4- 4, 4- 4)    panel-1                 gTree[panel-1.gTree.53]
## 3   2 ( 4- 4, 7- 7)    panel-2                 gTree[panel-2.gTree.68]
## 4   3 ( 4- 4,10-10)    panel-3                 gTree[panel-3.gTree.83]
## 5   4 ( 4- 4,13-13)    panel-4                 gTree[panel-4.gTree.98]
## 6   5 ( 4- 4,16-16)    panel-5                gTree[panel-5.gTree.113]
## 7   6 ( 4- 4,19-19)    panel-6                gTree[panel-6.gTree.128]
## 8   7 ( 4- 4,22-22)    panel-7                gTree[panel-7.gTree.143]
## 9   8 ( 3- 3, 4- 4)  strip_t-1    absoluteGrob[strip.absoluteGrob.211]
## 10  9 ( 3- 3, 7- 7)  strip_t-2    absoluteGrob[strip.absoluteGrob.217]
## 11 10 ( 3- 3,10-10)  strip_t-3    absoluteGrob[strip.absoluteGrob.223]
## 12 11 ( 3- 3,13-13)  strip_t-4    absoluteGrob[strip.absoluteGrob.229]
## 13 12 ( 3- 3,16-16)  strip_t-5    absoluteGrob[strip.absoluteGrob.235]
## 14 13 ( 3- 3,19-19)  strip_t-6    absoluteGrob[strip.absoluteGrob.241]
## 15 14 ( 3- 3,22-22)  strip_t-7    absoluteGrob[strip.absoluteGrob.247]
## 16 15 ( 4- 4, 3- 3)   axis_l-1 absoluteGrob[axis-l-1.absoluteGrob.199]
## 17 16 ( 4- 4, 6- 6)   axis_l-2         zeroGrob[axis-l-2.zeroGrob.200]
## 18 17 ( 4- 4, 9- 9)   axis_l-3         zeroGrob[axis-l-3.zeroGrob.201]
## 19 18 ( 4- 4,12-12)   axis_l-4         zeroGrob[axis-l-4.zeroGrob.202]
## 20 19 ( 4- 4,15-15)   axis_l-5         zeroGrob[axis-l-5.zeroGrob.203]
## 21 20 ( 4- 4,18-18)   axis_l-6         zeroGrob[axis-l-6.zeroGrob.204]
## 22 21 ( 4- 4,21-21)   axis_l-7         zeroGrob[axis-l-7.zeroGrob.205]
## 23 22 ( 5- 5, 4- 4)   axis_b-1 absoluteGrob[axis-b-1.absoluteGrob.150]
## 24 23 ( 5- 5, 7- 7)   axis_b-2 absoluteGrob[axis-b-2.absoluteGrob.157]
## 25 24 ( 5- 5,10-10)   axis_b-3 absoluteGrob[axis-b-3.absoluteGrob.164]
## 26 25 ( 5- 5,13-13)   axis_b-4 absoluteGrob[axis-b-4.absoluteGrob.171]
## 27 26 ( 5- 5,16-16)   axis_b-5 absoluteGrob[axis-b-5.absoluteGrob.178]
## 28 27 ( 5- 5,19-19)   axis_b-6 absoluteGrob[axis-b-6.absoluteGrob.185]
## 29 28 ( 5- 5,22-22)   axis_b-7 absoluteGrob[axis-b-7.absoluteGrob.192]
## 30 29 ( 7- 7, 4-22)       xlab             text[axis.title.x.text.249]
## 31 30 ( 4- 4, 2- 2)       ylab             text[axis.title.y.text.251]
## 32 31 ( 4- 4,24-24)  guide-box                       gtable[guide-box]
## 33 32 ( 2- 2, 4-22)      title               text[plot.title.text.261]

My negating-look-ahead-regex-fu is not working this morning, so if anyone with a shorter regex for this could edit it or comment that'd be awesome. Basically, we're filtering out the x-axis elements you don't want (print it again locally to see what's gone).

plot_filtered <- gtable_filter(plot_tab, 
                     "(background|panel|strip_t|axis_l|xlab|ylab|guide-box|title|axis_b-[1357])",
                     trim=FALSE)

And, now we do the actual plotting:

grid.newpage()
grid.draw(plot_filtered)

回答2:

This answer has gtable_filter_remove() function for simple negative subsetting. Based on the output from p_tab$layout$name, we would have to remove axis-b-2-1, axis-b-4-1 and axis-b-6-1 to get what we want.

library(ggplot2)

d <- ggplot(diamonds, aes(carat, price, fill = ..density..)) +
  xlim(0, 2) + stat_binhex(na.rm = TRUE) + theme(aspect.ratio = 1)

my_plot <- d + facet_wrap(~ color, nrow = 1)

library(gtable)
p_tab <- ggplotGrob(my_plot)
p_tab$layout$name

#>  [1] "background"  "panel-1-1"   "panel-2-1"   "panel-3-1"   "panel-4-1"  
#>  [6] "panel-5-1"   "panel-6-1"   "panel-7-1"   "axis-t-1-1"  "axis-t-2-1" 
#> [11] "axis-t-3-1"  "axis-t-4-1"  "axis-t-5-1"  "axis-t-6-1"  "axis-t-7-1" 
#> [16] "axis-b-1-1"  "axis-b-2-1"  "axis-b-3-1"  "axis-b-4-1"  "axis-b-5-1" 
#> [21] "axis-b-6-1"  "axis-b-7-1"  "axis-l-1-7"  "axis-l-1-6"  "axis-l-1-5" 
#> [26] "axis-l-1-4"  "axis-l-1-3"  "axis-l-1-2"  "axis-l-1-1"  "axis-r-1-7" 
#> [31] "axis-r-1-6"  "axis-r-1-5"  "axis-r-1-4"  "axis-r-1-3"  "axis-r-1-2" 
#> [36] "axis-r-1-1"  "strip-t-1-1" "strip-t-2-1" "strip-t-3-1" "strip-t-4-1"
#> [41] "strip-t-5-1" "strip-t-6-1" "strip-t-7-1" "xlab-t"      "xlab-b"     
#> [46] "ylab-l"      "ylab-r"      "guide-box"   "subtitle"    "title"      
#> [51] "caption"     "tag"

gtable_filter_remove <- function (x, name, trim = TRUE){
  matches <- !(x$layout$name %in% name)
  x$layout <- x$layout[matches, , drop = FALSE]
  x$grobs <- x$grobs[matches]
  if (trim) 
    x <- gtable_trim(x)
  x
}

p_filtered <- gtable_filter_remove(p_tab, name = paste0("axis-b-", c(2, 4, 6), "-1"),
                                   trim = FALSE)

library(grid)
grid.newpage()
grid.draw(p_filtered)

Created on 2018-08-26 by the reprex package (v0.2.0.9000).

回答3:

Is there a greater reason, why you decided to only display the plots in one row? Just set the parameter nrow = 2 of facet_wrap() and turn the x-axis text: change theme(aspect.ratio=1) to theme(aspect.ratio = 1,axis.text.x = element_text(angle = 90, hjust = 0,vjust=.5))

So the result would look like this:

library(ggplot2)
d <- ggplot(diamonds, aes(carat, price, fill = ..density..)) +
xlim(0, 2) + stat_binhex(na.rm = TRUE) +
theme(aspect.ratio = 1,axis.text.x = element_text(angle = 90, hjust = 0,vjust=.5))
d + facet_wrap(~ color, nrow = 2)

I know that might not answer your Question excactly, but for my understanding I thought you might be looking forward to make the plot less overloaded.

Does this work for you?