Non standard evaluation, lapply, and ggplot

Question

I'm trying to programmatically plot distributions using ggplot2.

I can't figure out how to work with non standard evaluation (NSE) here (even after having read Hadley's book chapter etc. on NSE).

Consider the following code:

library(ggplot2)


gg_dens <- function(x){
  eval(ggplot(data.frame(x), aes_string(x = substitute(x))) + geom_density() +
         ggtitle(substitute(x)), data.frame(x))
}


lapply(mtcars, function(x) gg_dens(x))

This code does produce a number of density plots, one for each column, ok. However, it does not print the name of the variable being plotted. Rather, the placeholder variable x is printed (cf. figure).

My goal is to have the x quote substituted by the real variable name, eg., mpg.

Answer 1

lapply is not going to work with the function you have right now to solve this. x is simply a vector when passed to that function, it is not the name of that variable, and lapply isn't passing anything that is the name. In other words, there is nothing in the scope of that function for it to figure out what should be the proper x-axis label.

One solution is similar to @Jimbou:

gg_dens <- function(name, dat) {
  ggplot(dat, aes_string(x = name)) + geom_density() + ggtitle(name)
}    
lapply(names(mtcars), gg_dens, mtcars)

Or just use facets instead:

mtcars2 <- tidyr::gather(mtcars)
ggplot(mtcars2, aes(value)) + geom_density() + facet_wrap(~key, scales = 'free')