How to increase the space between the bars in a bar plot in ggplot2?

Question

How can I increase the space between the bars in a bar plot in ggplot2 ?

Quillen

could you please provide code example illustrating your problem? In my experience the defaults of ggplot2 are really good, so there should be a good reason to change them. – Calvo

Maybe of interest for your specific question? :) #6085738 – Semination

Answer 1

You can always play with the width parameter, as shown below:

df <- data.frame(x=factor(LETTERS[1:4]), y=sample(1:100, 4))
library(ggplot2)
ggplot(data=df, aes(x=x, y=y, width=.5)) + 
  geom_bar(stat="identity", position="identity") +
  opts(title="width = .5") + labs(x="", y="") +
  theme_bw()

Compare with the following other settings for width:

So far, so good. Now, suppose we have two factors. In case you would like to play with evenly spaced juxtaposed bars (like when using space together with beside=TRUE in barplot()), it's not so easy using geom_bar(position="dodge"): you can change bar width, but not add space in between adjacent bars (and I didn't find a convenient solution on Google). I ended up with something like that:

df <- data.frame(g=gl(2, 1, labels=letters[1:2]), y=sample(1:100, 4))
x.seq <- c(1,2,4,5)
ggplot(data=transform(df, x=x.seq), aes(x=x, y=y, width=.85)) +
  geom_bar(stat="identity", aes(fill=g)) + labs(x="", y="") + 
  scale_x_discrete(breaks = NA) + 
  geom_text(aes(x=c(sum(x.seq[1:2])/2, sum(x.seq[3:4])/2), y=0, 
                label=c("X","Y")), vjust=1.2, size=8)

The vector used for the $x$-axis is "injected" in the data.frame, so that so you change the outer spacing if you want, while width allows to control for inner spacing. Labels for the $x$-axis might be enhanced by using scale_x_discrete().

Aideaidedecamp

Apparently, opts is no longer existing in R 4.0 and above. – Bollard

As of ggplot 3.0.0, you can use position_dodge2. It takes a padding argument that adds space in between the bars. See my answer below. – Loring

Answer 2

For space between factor bars use

ggplot(data = d, aes(x=X, y=Y, fill=F)) 
 + geom_bar(width = 0.8, position = position_dodge(width = 0.9))

The width in geom_bar controls the bar width in relation to the x-axis while the width in position_dodge control the width of the space given to both bars also in relation to the x-axis. Play around with the width to find one that you like.

Amil

Answer 3

As of ggplot 3.0.0, you can use position_dodge2. It takes a padding argument that adds space in between the bars.

library(ggplot2) # version >= 3.0.0
set.seed(1530676800) # 
tib <- tibble(
  group=rep(LETTERS[1:4], each=3), 
  subgroup=rep(letters[24:26], 4),
  y=sample(1:100, 4*3)
)
ggplot(tib, aes(x=group, fill=subgroup, y=y)) + 
  geom_col(position=position_dodge2()) 

Default padding is 0.1. Since I know you want to see different padding options:

library(purrr)
library(cowplot)
theme_set(theme_cowplot())

n_col <- 2
bottom_row_starts <- 3
plots0 <- purrr::imap(0.1*c(1:4), \(pad, idx) {
  ret <- ggplot(tib, aes(x=group, fill=subgroup, y=y)) + 
    geom_col(position=position_dodge2(padding=pad)) +
    labs(subtitle=paste0("Padding = ", pad)) #+
    #theme(plot.margin = margin(6, 0, 6, 0))
  if (bottom_row_starts > idx) {
    ret <- ret + xlab(NULL)
  }
  if(1 != (idx %% n_col)) {
    ret <- ret + ylab(NULL)
  }
  return(ret)
})
leg <- cowplot::get_legend(plots0[[1]])

plots <- plots0 %>% purrr::map(\(pp) {
  pp + theme(legend.position="none")
})

cowplot::plot_grid(
  plotlist=plots,
  ncol=n_col
) %>% cowplot::plot_grid(
  leg, 
  rel_widths = c(4, 1)
)

Loring

Answer 4

Thank you very much chl.! I had the same problem and you helped me solving it. Instead of using geom_text to add the X-labels I used scale_x_continuous (see below)

geom_text(aes(x=c(sum(x.seq[1:2])/2, sum(x.seq[3:4])/2), y=0, 
            label=c("X","Y")), vjust=1.2, size=8)

replaced by

scale_x_continuous(breaks=c(mean(x.seq[1:2]), mean(x.seq[3:4])), labels=c("X", "Y")) 

Ludivinaludlew

Answer 5

For space between POSIXlt bars you need adjust the width from the number of seconds in a day

# POSIXlt example: full & half width
d <- data.frame(dates = strptime(paste(2016, "01", 1:10, sep = "-"), "%Y-%m-%d"),
            values = 1:10)

ggplot(d, aes(dates, values)) +
  geom_bar(stat = "identity", width = 60*60*24) 

ggplot(d, aes(dates, values)) +
  geom_bar(stat = "identity", width = 60*60*24*0.5) 

Blodget