How to plot multiple factor columns with ggplot?

Question

I am trying to plot multiple factor columns in one go, using ggplot2 as plotting engine.

Plotting multiple metric columns is straight forward:

library(ggplot2)
library(dplyr) # dplyr 0.5.0 select_if
library(purrr)
data(diamonds)


diamonds %>% 
  select_if(is.numeric) %>% 
  gather %>% 
  ggplot(aes(x = value)) + 
  geom_histogram() + 
  facet_wrap(~key)

However, I did not success in plotting multiple factor (qualitative) columns in one shot. I would like to choose columns programmatically, ie., not directly naming them.

I tried this, but it does not produce a sensible result:

diamonds %>% 
  select_if(is.factor) %>% 
  gather %>% 
  ggplot(aes(x = value)) + geom_bar() +
  facet_wrap(~key) +
  coord_flip()

I assume that there might a solution along these lines:

diamonds %>% 
  select_if(is.factor) %>% 
  ggplot(aes(x = .[[1]])) + geom_bar()

Where .[[1]] should be replaced by some column placeholder (so here I directly named the column, which I would like to avoid, as I have a large number of columns in reality).

A for-loop will probably do the job, but I would like to get there with dplyr.

Could you explain what you would consider a sensible result, i.e. how your second figure differs from your desired output? Is the main problem the the fact that each plot doesn't have its own Y-axis (and therefore the alphabetical sorting is pretty meaningless)? — mkt
– mkt, Commented Jul 4, 2016 at 12:16
#mkt, thanks for your comment, #Axeman had a good idea, that was what I was looking for — Sebastian Sauer
– Sebastian Sauer, Commented Jul 4, 2016 at 17:57

Axeman · Accepted Answer · 2016-07-04 12:33:02Z

3

The trick here is to use scales = free within your facet call. For example:

diamonds %>% 
    select_if(is.factor) %>% 
    gather %>% 
    ggplot(aes(x = value)) + geom_bar() +
    facet_wrap(~key, scales = 'free') +
    theme_bw() +
    theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5))

Unfortunately free scales and coord_flip don't play nice. You can either use geom_barh from the ggstance package. Or you can use lapply on each column to get a list of ggplot objects and use the cowplot package to stitch them together in one figure.

answered Jul 4, 2016 at 12:33

Axeman

35.7k8 gold badges87 silver badges101 bronze badges

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

1 Comment

Sebastian Sauer Over a year ago

It just came to my mind that lapply might also be a way, maybe easier to play together with coord_flip@Axeman

Sebastian Sauer · Accepted Answer · 2016-07-04 17:56:12Z

1

Maybe less elegant than @Axeman, but also working, and cooperates with coord_flip:

library(gridExtra)


gg_bar <- function(x, ...){
  { 
    ggplot(data_frame(x), aes(x = x)) + 
      geom_bar() +
      coord_flip() 
  }
}

diamonds %>% 
  select_if(negate(is.numeric)) %>% 
  lapply(., function(x) gg_bar(x)) -> gg_bar_list

do.call(grid.arrange, gg_bar_list)

However, the name of the variable ("x") is not shown, that's not too beautiful.

answered Jul 4, 2016 at 17:56

Sebastian Sauer

1,6931 gold badge17 silver badges25 bronze badges

1 Comment

Axeman Over a year ago

See cowplot::plot_grid instead of grid.arrange to do automatic alignment and subplot annotation. You can use the annotations as variable labels, and not show the y-axis label.

Collectives™ on Stack Overflow

How to plot multiple factor columns with ggplot?

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related