Side-by-Side bar plots with consistent colour scheme

Question

I have been using a consistent colour scheme across multiple bar plots using data from different data frames that share a common variable. This approach was based on an answer to this question. In each of these cases I was looking at one variable.

What I would like to do now is look at two variables on the same plot keeping the colour scheme consistent. It should look something like the following except the two bars for A should be the same colour (the first in the palette), the two bars for B should be the same colour (the second in the palette), etc.

Sample data:

mydata <- data.frame( "Channel" = LETTERS[1:11], 
                  "V1" = seq(1100, 100, -100), 
                  "V2"= seq(110, 10, -10) )

and code for the above plot:

barplot(t(mydata[,2:3]), beside=T, 
    col=brewer.pal(11, "RdGy"), 
    names.arg=LETTERS[1:11])

I have been trying to do this in ggplot (which is my preference) but haven't succeeded. I have used the package reshape2 to melt the data and then use fill but that only gets me two colours. I have also tried using position='dodge' in various ways with no luck. I can get to a similar end result combining a bar plot and a line graph for the second variable but would prefer the double bar plot.

I would also prefer to stick with ggplot but any solution would be appreciated.

The example code gives error `'height' must be a vector or a matrix` — dww, Jun 29 '16 at 20:59
@dww Sorry and thanks for catching. I was stuck in `ggplot2` mode and didn't convert argument to matrix. — pentandrous, Jun 29 '16 at 21:23

score 2 · Accepted Answer · edited Nov 26 '21 at 12:36

I solved it using a little hack. I set the color aesthetic to Channel and family to variables. The second one prevents that both values with the same Channel are plotted on top of each other when using position = "dodge".

library(ggplot2)
library(reshape2)
library(dplyr)

mydata <- data.frame( "Channel" = LETTERS[1:11], 
                      "V1" = seq(1100, 100, -100), 
                      "V2"= seq(110, 10, -10) )

df <- mydata %>% 
      melt(id = c("Channel"))

ggplot(df, aes(x = Channel, y = value, fill = Channel, family = variable)) + 
  geom_bar(stat = "identity", position = "dodge", color = "black") + 
  scale_fill_brewer(palette = "RdGy")

If you do not like the background you could use another theme. For example theme_base() from the ggthemes package should look similar to your barplot,

Awesome, thanks! Just what I needed. And `ggplot` was the approach I was hoping for. — pentandrous, Jun 29 '16 at 23:01

dww · Answer 2 · 2016-06-29T23:32:44.337

1

You can use base graphics by specifying the colours to be repeated for each adjacent bar.

barplot(t(mydata[,2:3]), beside=T, col=rep(brewer.pal(11, "RdGy"), each=2), names.arg=LETTERS[1:11])

edited Jun 29 '16 at 23:32

answered Jun 29 '16 at 23:24

dww

30,425
5
68
111

Thanks! Always good to have more than one option and this code is a bit more straightforward. – pentandrous Jun 30 '16 at 11:33

Side-by-Side bar plots with consistent colour scheme

2 Answers2