R Plotting Series of Data with Secondary axis

Question

I would like to plot a data frame with 4 columns:

Quarters <-  c("Q1","Q2","Q3")
Series1 <- c("1%","2%","3%")
Series2 <- c("4%","5%","6%")
Series3 <- c("1000","2000","3000")

df <- data.frame(Quarters,Series1,Series2,Series3)

Quarters as x-axis, Series1 & Series2 as left y-axis, Series3 as right y-axis and a legend.

I have seen some solutions with ggplot using scale_y_continues, but then the secondary (y) axis has to be a multiple of the primary axis. Which I do not want, as the data will be dynamic and the ratio might not hold through in all instances.

Any solutions how I might go about creating this? Perhaps ggplot is not the way to go?

You can add a secondary axis that is a transformation of the primary axis with `sec_axis`: http://ggplot2.tidyverse.org/reference/sec_axis.html But ggplot is philosophically opposed to multiple unrelated axes, so if that's *really* what you want you're going to have to look elsewhere: https://stackoverflow.com/a/3101876 — divibisan, May 24 '18 at 14:24

score 1 · Accepted Answer · edited May 24 '18 at 18:58

I don't know about ggplot2, but you can use par(new = T) in R to plot a graph on top of another one. If you remove the right axis from the first plot and add it manually on the second one it should look good.

Quarters <-  c(1,2,3)
Series1 <- c(0.01,0.02,0.03)
Series2 <- c(0.04,0.05,0.06)
Series3 <- c(1000,2000,3000)

par(mar = c(5,5,2,5)) # Leaves some space for the second axis

plot(Quarters,Series1,type="l",ylim=c(0,0.1))
lines(Quarters,Series2,col="red")

par(new=T)
plot(Quarters,Series3,type="l",axes=F, xlab=NA, ylab=NA,col="blue") # Removes axis and labels so they don't overlap
axis(side = 4) # Adds secondary axis

Does this work for you? More info here

Yes, this definitely answers my question. Thank you. – Aksel Etingu May 25 '18 at 07:29 — Aksel Etingu, May 25 '18 at 07:29

score 1 · Answer 2 · edited Jun 20 '20 at 09:12

ggplot2 is perfectly fine and deals with dual-axis very well. You would use sec.axis within scale_y_continuous or scale_y_discrete (or really just about any valid scale_y_) call:

  scale_y_continuous(
    "Casualties* due to:",
    sec.axis = sec_axis(~. *0.001, 
                        name="Aircraft passengers carried, bn", 
                        labels = scaleFUN, 
                        breaks = seq(0,3, by=0.5)),
    limits = c(0,3000),
    breaks = seq(0,3000, by=500),
    labels = comma
  )

The following creates two axis, one with a break of 0 to 3000, by 500. That's the axis on the left (primary axis). The second one goes by 0 to 3 by 0.5, but there's no reason why it should follow that scale. You can very well have scales that are not multiples of the primary axis.

You can get a plot like the following:

Using the above technique. If it is helpful I put up the full ggplot code to recreate the above plot in this post. Completely done in ggplot2 including the horizontal legend and secondary axis.

Good luck!

Thank you very much. Although @Fino has given me what I was looking for, I have to admit ggplot2 is a lot snappier. — Aksel Etingu, May 25 '18 at 07:32
I'm not sure if this is correct: the [discrete axes](https://www.rdocumentation.org/packages/ggplot2/versions/3.2.1/topics/scale_x_discrete) do not support the `sec.axis` attribute! — Yan Foto, Jan 31 '20 at 10:07

R Plotting Series of Data with Secondary axis

2 Answers2