Bar plot, no space between bottom of geom and x axis keep space above

Question

When I plot a bar graph in ggplot2 I would like to reduce the space between the bottom of the bars and the x-axis to 0, yet keep the space above the bars and the plot box. I have a hack to do it below. It's dirty and I want to be clean again. Is there a way to achieve this behavior without the dirty little hack?

Default (desired space above but don't want space below bars):

ggplot(mtcars, aes(x=as.factor(carb))) + 
    geom_bar()

enter image description here

Use expand (undesired 0 space above but got the 0 space below bars):

ggplot(mtcars, aes(x=as.factor(carb))) + 
    geom_bar() + 
    scale_y_continuous(expand = c(0,0))

enter image description here

Dirty Hack (I like it but its.. well, dirty):

ggplot(mtcars, aes(x=as.factor(carb))) + 
    geom_bar() + 
    scale_y_continuous(expand = c(0,0)) +
    geom_text(aes(x=1, y=10.3, label="Stretch it"), vjust=-1)

enter image description here

I assume you consider using `coord_cartesian` to much hardcoding as well? — Henrik, Nov 26 '13 at 15:08
@baptise can you add as a solution for future searchers. That works as well. — Tyler Rinker, Nov 26 '13 at 15:12
Good question, this also annoys me in ggplot2 graphs that y axis does not start from the bottom of the picture frame. — jrara, Nov 27 '13 at 11:41
There's a new(er) question that points back to this one, with a [really great solution](http://stackoverflow.com/q/22480052/903061) generalizing the `expand()` into a list of upper limit expansion and lower limit expansion. — Gregor Thomas, Dec 17 '15 at 21:10

score 49 · Answer 1 · edited Dec 24 '15 at 17:12

49

I might be missing what you really want, but without using geom_text hack you can still set the limits

ggplot(mtcars, aes(x = as.factor(carb))) + 
    geom_bar() + 
    scale_y_continuous(expand = c(0, 0), limits = c(0, 10.3)) 

# marginally cleaner

edited Dec 24 '15 at 17:12

Tyler Rinker

108,132
65
322
519

answered Nov 26 '13 at 15:06

user1317221_G

15,087
3
52
78

1

Yes exactly what I wanted. I tried to use: `ylim(c(0,10.3))` after the `scale_y_continuous` and it was a no go. I get it now. Pretty simple. Thanks. – Tyler Rinker Nov 26 '13 at 15:09
1

I removed the check and awarded to a better answer at this time. I awared a bounty to your answer so no points were lost. Thank you. Your original solution was very useful until it was addressed by the ggplot2 team. – Tyler Rinker Aug 27 '20 at 19:10

bee5911 · Accepted Answer · 2020-08-23T06:35:32.123

39

The R documentation includes a new convenience function called expansion for the expand argument as the expand_scale() became deprecated as of ggplot2 v3.3.0 release.

ggplot(mtcars) +
  geom_bar(aes(x = factor(carb))) + 
  scale_y_continuous(expand = expansion(mult = c(0, .1)))

edited Aug 23 '20 at 06:35

answered Aug 23 '20 at 06:28

bee5911

516
4
9

2

This is the correct answer and should receive the checkmark. – Claus Wilke Aug 26 '20 at 00:44
The trick is to use the expansion combined .1 which is not said by many answers that I've lookup – testing_22 Apr 29 '21 at 01:43

score 19 · Answer 3 · answered Nov 26 '13 at 15:24

19

You can expand the limits manually, e.g. with expand_limits(y=10.1), or use this trick to add an invisible layer with scaled up data,

ggplot(mtcars, aes(x=as.factor(carb))) + 
    geom_bar() + 
    scale_y_continuous(expand = c(0,0)) +
    geom_blank(aes(y=1.1*..count..), stat="bin")

answered Nov 26 '13 at 15:24

baptiste

75,767
19
198
294

1

Ahaa, those `..` variables...+1! – Henrik Nov 26 '13 at 15:32
Is there a way to do this without hard coding (1.1)? – jrara Nov 27 '13 at 11:43
I mean that this multiplication by 1.1 works in this case but if you have dynamically changing y-axis and you do not know the scale beforehand. In these kind of cases you somehow have to use a parameter instead of hardcoding this multiplication factor. It would be nice if ggplot could do this automatically based on data (y axis scale). – jrara Nov 27 '13 at 14:04
2

i still don't understand - the `..count..` variable takes care of this automatic adjustment of the y-axis, but at some point it is up to the user to define what margin should be added above the bar (here 10% of the full data range) and that's what this multiplication factor does. It's the same with default axes in base R and ggplot2. – baptiste Nov 27 '13 at 15:42

score 15 · Answer 4 · answered Sep 13 '18 at 17:02

Starting in ggplot2 3.0.0 there is an expand_scale() function that can be used with the expand argument to do exactly this. You define the top and bottom expansion separately.

ggplot(mtcars, aes(x=factor(carb))) + 
     geom_bar() +
     scale_y_continuous(expand = expand_scale(mult = c(0, .1)))

score 9 · Answer 5 · answered Nov 26 '13 at 15:16

9

Because you seem comfortable with some hardcoding...

ggplot(mtcars, aes(x = as.factor(carb))) + 
  geom_bar() +
  coord_cartesian(ylim = c(0, 10.3))

answered Nov 26 '13 at 15:16

Henrik

65,555
14
143
159

I didn't realize before how powerful this answer can be and generalizable to categorical scales as well. – Tyler Rinker Sep 11 '14 at 18:56

score 3 · Answer 6 · answered Jun 05 '18 at 09:51

This is an automatic way to produce the spacing at the top, yet remove the bottom spacing. I use 3 % padding since that's what you hard-coded.

plot1 <- ggplot(mtcars, aes(x=as.factor(carb))) +
    geom_bar()

plotInfo <- print(plot1)
yMax <- max(plotInfo$data[[1]]$ymax)
yLimitMax <- 1.03 * yMax

plot2 <- plot1 +
    scale_y_continuous(expand = c(0,0),
                       limits = c(0,yLimitMax))

If you want to remove the three lines between the plots, just write this in plot2 instead:

limits = c(0, 1.03 * max(print(plot1)$data[[1]]$ymax))

Bar plot, no space between bottom of geom and x axis keep space above

6 Answers6

Linked

Related