Putting x-axis at top of ggplot2 chart

Question

I feel like this should be obvious... all I'm trying to do is to remove the x-axis from the bottom of my graph and add it to the top.

Here is a reproducible example. Data plus code to make the following graph:

library(reshape2)
library(ggplot2)

data(mtcars)
dat <- with(mtcars, data.frame(mpg, cyl, disp, hp, wt, gear))
cor.matrix <- round(cor(dat, use = "pairwise.complete.obs", method = "spearman"), digits = 2)
diag(cor.matrix)<-NA

cor.dat <- melt(cor.matrix)
cor.dat <- data.frame(cor.dat)
cor.dat <- cor.dat[complete.cases(cor.dat),]


ggplot(cor.dat, aes(Var2, Var1, fill = value)) + 
  geom_tile(colour="gray90", size=1.5, stat="identity") + 
  geom_text(data=cor.dat, aes(Var2, Var1, label = value), color="black", size=rel(4.5)) +
  scale_fill_gradient(low = "white", high = "dodgerblue", space = "Lab", na.value = "gray90", guide = "colourbar") +
  scale_x_discrete(expand = c(0, 0)) +
  scale_y_discrete(expand = c(0, 0)) +
  xlab("") + 
  ylab("") +
  theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_rect(fill=NA,color="gray90", size=0.5, linetype="solid"),
        axis.line = element_blank(),
        axis.ticks = element_blank(),
        panel.background = element_rect(fill="gray90"),
        plot.background = element_rect(fill="gray90"),
        legend.position = "none", 
        axis.text = element_text(color="black", size=14) )

enter image description here

But what I'm trying to produce is- it seems like this should be obvious (e.g. it's very easy to do in base-R) , but I haven't managed to find what I'm looking for in ggplot2.

enter image description here

The axes in ggplot aren't nearly as flexible as in base R. For your example, you're not using any axis components except for labels. I'd say your best bet is to not plot an axis at all, and just use `geom_text` to put the labels where you want. — Gregor Thomas, Nov 10 '14 at 06:46
Moving the x-axis is difficult, but it is possible by using the functions ggplot_gtable and ggplot_build. These functions "plot" the ggplot, but instead of to the screen, makes socalled "grob"-objects that represent the visual elements (a box, line, grid, etc.), but on a different level than the ggplot objects (scales, themes, aestetics, etc.). You can manipulate the grobs as you see fit, swapping them, resizing them etc. — MrGumble, Nov 10 '14 at 07:32
Unless you're automating this, you may just want to make the figure then edit it after the fact. Would take a few seconds to move the axis in illustrator or inkscape. — Minnow, Nov 29 '14 at 02:22
I found this page since I am working on a Shiny app; inkscape or Illustrator is out of the question! — tumultous_rooster, Dec 05 '14 at 17:40

score 169 · Accepted Answer · answered Jan 11 '17 at 18:51

169

You can move the x-axis labels to the top by adding

scale_x_discrete(position = "top")

answered Jan 11 '17 at 18:51

Matthias Haber

1,706
1
9
2

Note that this solution may not work as expected if you are using `coord_flip()` in your code. I think this is larger issue in ggplot2 where the 'x' and 'y' parameters of some of the functions like `scale...` and `labs` continue to mean 'x' and 'y' as defined in the aes() call and others like 'facet_wrap' switch their meaning when you flip the coordinates. You can work around this easily, by manually flipping your x and y variables in the aes() call. – ESELIA Feb 24 '22 at 14:27

score 6 · Answer 2 · answered Jun 13 '15 at 04:20

I've used this work around. Used a duplicated the plot's x-axis and affixed the two together, then crop. I don't have the cleaned up code, but below is an example.

p.bot   <-  
ggplot(cor.dat, aes(Var2, Var1, fill = value)) + 
  geom_tile(colour="gray90", size=1.5, stat="identity") + 
  geom_text(data=cor.dat, aes(Var2, Var1, label = value), color="black", size=rel(4.5)) +
  scale_fill_gradient(low = "white", high = "dodgerblue", space = "Lab", na.value = "gray90", guide = "colourbar") +
  scale_x_discrete(expand = c(0, 0)) +
  scale_y_discrete(expand = c(0, 0)) +
  xlab("") + 
  ylab("") +
  theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_rect(fill=NA,color="gray90", size=0.5, linetype="solid"),
        axis.line = element_blank(),
        axis.ticks = element_blank(),
        panel.background = element_rect(fill="gray90"),
        plot.background = element_rect(fill="gray90"),
        legend.position = "none", 
        axis.text.x = element_blank(),
        plot.margin = unit(c(1,0,0,0), "cm"),
        axis.text.y = element_text(color="black", size=14) )

p.top   <-  p.bot + theme(
    axis.text.x = element_text(color="black", size=14),
    axis.text.y = element_text(color="gray90", size=14)
    )  + coord_cartesian(ylim = c(0,0))



require(gtable)
#Extract Grobs
g1<-ggplotGrob(p.top)
g2<-ggplotGrob(p.bot)
#Bind the tables
g<-gtable:::rbind_gtable(g1, g2, "first")
#Remove a row between the plots
g <- gtable_add_rows(g, unit(-1.25,"cm"), pos=nrow(g1))
#draw
panels <- g$layout$t[grep("panel", g$layout$name)]
g$heights[panels] <- lapply(c(0,2), unit, "null")

grid.newpage()
grid.draw(g)

enter image description here

score 5 · Answer 3 · answered Feb 11 '16 at 04:40

5

check out the cowplot package

ggdraw(switch_axis_position(p + axis = 'x'))

answered Feb 11 '16 at 04:40

Jerry T

1,541
1
19
17

Under the current version of cowplot, I needed to swap the plus for a comma: ggdraw(switch_axis_position(p, axis = 'x')) – Jessica Beyer Sep 12 '16 at 14:51

score 3 · Answer 4 · answered May 29 '21 at 04:06

Nowadays, you can do this by adding "position = 'top'" in the scale_x_discrete.

code:

library(reshape2)
library(ggplot2)

data(mtcars)
dat <- with(mtcars, data.frame(mpg, cyl, disp, hp, wt, gear))
cor.matrix <- round(cor(dat, use = "pairwise.complete.obs", method = "spearman"), digits = 2)
diag(cor.matrix)<-NA

cor.dat <- melt(cor.matrix)
cor.dat <- data.frame(cor.dat)
cor.dat <- cor.dat[complete.cases(cor.dat),]


ggplot(cor.dat, aes(Var2, Var1, fill = value)) + 
  geom_tile(colour="gray90", size=1.5, stat="identity") + 
  geom_text(data=cor.dat, aes(Var2, Var1, label = value), color="black", size=rel(4.5)) +
  scale_fill_gradient(low = "white", high = "dodgerblue", space = "Lab", na.value = "gray90", guide = "colourbar") +
  scale_x_discrete(expand = c(0, 0),position = 'top') +
  scale_y_discrete(expand = c(0, 0),position = 'left') +
  xlab("") + 
  ylab("") +
  theme(panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_rect(fill=NA,color="gray90", size=0.5, linetype="solid"),
        axis.line = element_blank(),
        axis.ticks = element_blank(),
        panel.background = element_rect(fill="gray90"),
        plot.background = element_rect(fill="gray90"),
        legend.position = "none", 
        axis.text = element_text(color="black", size=14) )

This is the same answer as the accepted one. – ESELIA Feb 24 '22 at 14:23 — ESELIA, Feb 24 '22 at 14:23

score 1 · Answer 5 · edited Jan 08 '18 at 22:07

1

see http://ggplot2.tidyverse.org/reference/sec_axis.html

scale_y_continuous(sec.axis = dup_axis())

edited Jan 08 '18 at 22:07

Floern

33,559
24
104
119

answered Jan 08 '18 at 20:28

smroecker

11
1

2

Consider expanding your answer by explaining what the code does/why it solves the OP's problem and summarizing the details that are available at the link your shared. – Joey Harwood Jan 08 '18 at 20:48
1

While this code snippet may be the solution, including an explanation really helps to improve the quality of your post. Remember that you are answering the question for readers in the future, and those people might not know the reasons for your code suggestion – Rahul Gupta Jan 09 '18 at 05:26

score -1 · Answer 6 · answered Jan 27 '23 at 00:43

-1

if it is a continuous scale use

scale_x_continuous(position = "top") +

if it is a discrete scale use

scale_x_discrete(position = "top") +

answered Jan 27 '23 at 00:43

Caleb

1

2

How is this answer different from some of the others? Also the use of the `+` operator here is a bit confusing. I would either provide the full code chunk or omit this. If for example you used this function at the end of a `ggplot` object, it would cause an error. – Shawn Hemelstrand Jan 27 '23 at 02:02

Putting x-axis at top of ggplot2 chart

6 Answers6

Linked