Pass interactions between variables to a function using dplyr/ggplot

Question

I often find it useful to write functions that create a plot using different subsets of a dataset. For example, take data like these:

id <- seq(1:640)
con1 <- rep(c('hi', 'lo'), each = 320, times = 1)
con2 <- rep(c('up', 'down'), each = 160, times = 2)
con3 <- rep(c('fork', 'knife'), each = 80, times = 4)
con4 <- rep(c('carrot', 'broccoli'), each = 40, times = 8)
likert <- sample(c(1:5), 640, replace = T)

dat <- as.data.frame(cbind(id, con1, con2, con3, con4, likert))

I would like to have a function that would create a facetted plot for any combination of the con1:4 variables that I specify. A simplified version of what I would like looks like this:

groupVar <- 'con1'

plotFunction <- function(groupVar) {

  plot <- ggplot(data = dat) +
    geom_bar(aes(x = factor(likert))) +
    facet_wrap(~eval(parse(text = groupVar))) +
    ggtitle(paste('Preferences grouped by ',groupVar)) + 
    theme_bw()

  return(plot)

}

plotFunction(groupVar = groupVar)

This works for passing single variables, but not when I want to plot an interaction. For example, setting groupVar to 'con1*con2' does not result in the desired result - ie, the plot I would get from manually specifying

facet_wrap(~con1*con2)

I understand (no doubt on a fairly superficial level) why this doesn't work, but haven't been able to figure out the solution. My workaround has been to create new variables for each interaction, so that I can then pass the interaction to facet_wrap as a single variable - eg for the con1*con2 interaction:

dat <- dat %>%
  mutate(con1_con2 = paste(con1, con2, sep = '_'))

groupVar <- 'con1_con2'

plotFunction(groupVar = groupVar) 

gives the desired result.

I would be grateful for any advice on how I can pass interactions to the function via the groupVar variable (or some other method) so that there is no need to create the new variables.

Answer 1

You can use the character vector argument syntax instead of the ~ formula syntax ( docs ):

groupVar <- c('con1','con2')

plotFunction <- function(groupVar) {

  plot <- ggplot(data = dat) +
    geom_bar(aes(x = factor(likert))) +
    facet_wrap(groupVar) +
    ggtitle(paste('Preferences grouped by ', paste0(groupVar, collapse="*"))) + 
    theme_bw()

  return(plot)

}

plotFunction(groupVar = groupVar)

Just remember to update the ggtitle paste command as well.