Create R boxplots of dataframe with another variable

Question

I'm looking to create a set of box plots where I create a bloxplot for each variable in sampledf1 against the single variable in sampledf2.

The actual use case is I've created a set of clusters with k-means and now want to see their distribution for each of the found clusters with each variable in the dataframe I'm using for clustering.

sampledf1 <- as.data.frame(replicate(6, sample(c(1:10,NA))))
sampledf2 <- as.data.frame(replicate(1, sample(c(21:30,NA))))

Then I want to see a box plot with each of the variables in sampledf1 paired with the only variable in sampledf2.

I would like to use something like:

sapply(boxplot(sampledf1~sampledf2$V1))

but this gives me this error:

Error in match.fun(FUN) : argument "FUN" is missing, with no default

Anyway I could do this would dplyr would be great but I didn't see any functions that I could chain together to do this.

bouncyball · Accepted Answer · 2016-12-27 18:06:51Z

3

Here's a way using lapply and seq_along. We iterate through the columns of sampledf1 using seq_along. We can extract the variable names using our index, i, and the names function.

par(mfrow = c(2,3))
lapply(seq_along(sampledf1), 
       FUN  = function(i) 
           boxplot(sampledf1[,i] ~ sampledf2$V1, main = names(sampledf1)[i])
       )

answered Dec 27, 2016 at 18:06

bouncyball

10.8k23 silver badges32 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

dww · Accepted Answer · 2016-12-27 18:21:59Z

2

You can use ggplot and facets, if you first reshape your data into long format

library(reshape2)
library(ggplot2)
s.all = cbind(sampledf1, f2=sampledf2$V1)
s.long = melt(s.all, id = 'f2')
ggplot(s.long) +
  geom_boxplot(aes(x=f2, group=f2, y=value)) +
  facet_wrap(~variable) +
  scale_x_continuous(breaks=unique(s.long$f2))

edited Dec 27, 2016 at 18:21

answered Dec 27, 2016 at 18:14

dww

31.6k8 gold badges75 silver badges126 bronze badges

1 Comment

Jazzmine Over a year ago

Thanks for sharing this as it's good to have more than one way to solve a problem and I'm learning plot.

Nate · Accepted Answer · 2016-12-27 18:21:29Z

1

library(purrr)'s walk works nicely when you start trying to pass formulas like this. walk() works like sapply, iterating over the elements in an object, just with more flexible syntax. The . refers to the iterated element from names(sampledf1).

This will work to get each panel named by the column in sampledf1 it represents:

library(purrr)    
par(mfrow = c(2,3))
purrr::walk(names(sampledf1), ~boxplot(sampledf1[,.]~sampledf2$V1, main = .))

edited Dec 27, 2016 at 18:21

answered Dec 27, 2016 at 17:56

Nate

10.7k3 gold badges35 silver badges42 bronze badges

2 Comments

Jazzmine Over a year ago

Thanks Nathan, is there anyway to add labels so I know which data each box plot has? Or is by order of variables in sampledf1 if I were to do a names(sampledf1)?

Nate Over a year ago

you're correct about the order it process in. you don't really need map here but it is the same function with different output, each useful depending on the scenario. I will edit in a labeled version

Thales · Accepted Answer · 2016-12-27 18:16:42Z

0

ggplot2 variant:

library(reshape2)
library(ggplot2)

sampledf1$X <- sampledf2$V1
ggplot(melt(sampledf1, id.vars="X", na.rm=T), aes(factor(X),value)) + 
  geom_boxplot() + facet_wrap( ~ variable, nrow=2)

answered Dec 27, 2016 at 18:16

Thales

6054 silver badges10 bronze badges

Collectives™ on Stack Overflow

Create R boxplots of dataframe with another variable

4 Answers 4

Comments

1 Comment

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

Comments

1 Comment

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related