How to overlay scatterplots in R?

Question

If I have three sets of data :

a1= rnorm(10)
a2= rnorm(10)
a3= rnorm(10)

rather than looking at these side by side using:

par(mfrow=c(1,3))
plot(a1)
plot(a2)
plot(a3)

How do I get all these points on the same plot?

In base graphics, you add things to an existing plot using functions like `points`, `lines`, etc., rather than calling `plot` again. (Note that this is demonstrated in the last example in `?plot`.) — joran, Oct 01 '13 at 16:08

csgillespie · Answer 1 · 2013-10-02T08:58:50.923

36

Just use the points function:

plot(a1)
points(a2, col=2)
points(a3, col=3)

This is equivilent to:

plot(1:length(a1), a1)
points(1:length(a2), a2, col=2)
points(1:length(a3), a3, col=3)

If the vectors have unequal lengths, then you should specify the x-axis limit:

plot(a1, xlim=c(1, max(length(a1), length(a2), length(a3))))

edited Oct 02 '13 at 08:58

answered Oct 01 '13 at 16:20

csgillespie

59,189
14
150
185

To accommodate for their additional query of how to deal with vectors of different lengths, this will need to be modified from a "just use" type of answer. – A5C1D2H2I1M1N2O1R2T1 Oct 02 '13 at 04:49
@AnandaMahto The OP solution doesn't actually match the question - they've swapped the x and y axis around. – csgillespie Oct 02 '13 at 08:55
I agree that the question was not really well-phrased. I think the OP was using SO a bit as a personal wiki and didn't really think through exactly what they wanted to ask :) – A5C1D2H2I1M1N2O1R2T1 Oct 02 '13 at 09:00

score 9 · Answer 2 · edited Jun 23 '15 at 19:29

# To overlay scatterplots in R

# import the required libraries
library(ggplot2)
library(reshape2)

# assign data
a1=rnorm(10)
a2=rnorm(10)
a3=rnorm(10)

# create a dataframe from combined data
# and set count to however many points are in each dataset
df = data.frame(a1, a2, a3, count = c(1:10))

# melt the dataframe
df.m = melt(df, id.vars ="count", measure.vars = c("a1","a2","a3"))

# take a look at what melt() does to get an idea of what is going on
df.m

# plot out the melted dataframe using ggplot
ggplot(df.m, aes(count, value, colour = variable)) + geom_point() + ylim(-3,3)

# swapping the axis
ggplot(df.m, aes(value, count, colour = variable)) + geom_point() + xlim(-3,3)

When a1 and a3 are equal in size, it is not possible to put into the same data.frame as columns, as input for melt. The solution is to simply use a list:

a1 = rnorm(10)
a2 = rnorm(25)
a3 = rnorm(17)
a_list = list(a1, a2, a3)
a_df = do.call("rbind", lapply(a_list, 
                                function(x) data.frame(value = x, 
                                                       count = seq_along(x))))
ID_options = LETTERS[seq_along(a_list)]
a_df$ID = rep(ID_options, sapply(a_list, length))
ggplot(a_df, aes(x = value, y = count, color = ID)) + geom_point()

enter image description here

@PaulHiemstra This plot isn't the same as `plot(a1)` The x and y axis have been swapped. — csgillespie, Oct 02 '13 at 08:55

score 3 · Answer 3 · edited May 23 '17 at 12:07

To add variety to the answers, you can also use lattice. Here, the second line in each set of code samples represent swapped axes.

library(lattice)

## If you have already created the "df" 
##   data.frame from your example
xyplot(count ~ a1 + a2 + a3, data=df)
xyplot(a1 + a2 + a3 ~ count, data=df)

## Without first creating the "df" 
##   data.frame from your example    
xyplot(1:10 ~ a1 + a2 + a3)
xyplot(a1 + a2 + a3 ~ 1:10)

If you are working with vectors of unequal lengths, you can load the functions from this answer I shared about cbinding vectors of unequal lengths and then use the first approach I mentioned. Update: See https://gist.github.com/mrdwab/6789277 for the most recent versions of the function.

Example:

a1 = rnorm(10)
a2 = rnorm(25)
a3 = rnorm(17)

library(lattice)
library(devtools)
## source_gist is not working properly unless you provide 
##   the full URL to the "raw" file
source_gist("https://gist.github.com/mrdwab/6789277/raw/9bd7d5931389ec475c49c1918d26d9899796a5d0/Cbind.R")

newdf <- Cbind(a1, a2, a3)
xyplot(a1 + a2 + a3 ~ sequence(nrow(newdf)), data=newdf)
xyplot(sequence(nrow(newdf)) ~ a1 + a2 + a3, data=newdf)

Here's an example plot with a little bit of adjustments to the default colors:

xyplot(sequence(nrow(newdf)) ~ a1 + a2 + a3, data=newdf, 
       pch = 21, fill = c("black", "red", "green"), cex = 1)

enter image description here

How to overlay scatterplots in R?

3 Answers3