How to add a second legend using ggpairs in R?

Question

Basically, I'm trying to add a second legend to a ggpairs plot. In the example plot below, I am using the iris data and running a randomForest to get the predictions. I am plotting the density on the diagonal, a scatterplot in the lower triangle and I'm plotting the correlation in the upper triangle. I am colouring the densities and scatterplot by the randomForest predictions. However, I am colouring the correlation grid using a custom function (found here).

I was wondering if it's possible to have a legend displaying the Species (this part is in the code below) and a legend showing the correlation colours. It is the latter part that I'm having trouble with.

library(randomForest)
library(GGally)

# Get data
ir <- iris

# Create randomForest
rf <- randomForest::randomForest(Species~., data = ir)
pred <- rf$predicted


# Remove Species for plotting
ir <- ir[,-5]

# function to colour corr plot
my_fn <- function(data, mapping, method="p", use="pairwise", ...){
  
  # grab data
  x <- eval_data_col(data, mapping$x)
  y <- eval_data_col(data, mapping$y)
  
  # calculate correlation
  corr <- cor(x, y, method=method, use=use)
  
  # calculate colour based on correlation value
  colFn <- colorRampPalette(c("dodgerblue4", "floralwhite", "firebrick1"), interpolate ='spline')
  fill <- colFn(100)[findInterval(corr, seq(-1, 1, length=100))]
  
  #make plot
  ggally_cor(data = data, mapping = mapping,  ...) +
    theme_void() +
    theme(panel.background = element_rect(fill=fill), panel.border=element_blank(), axis.line=element_line(),
          strip.text = element_text(face="bold", colour="red", size = 5))
}


  
p <- ggpairs(ir,
             mapping=ggplot2::aes(colour = pred), legend = 1,
             upper=list(continuous = my_fn),
             diag = list(continuous = "densityDiag"),
             lower=list(continuous=wrap("points", size=.2))) +
  theme(panel.border=element_blank(), axis.line=element_line(),
                   strip.text = element_text(face="bold", colour="red", size = 5))
p

This results in something that will look like this:

Is it possible to add a second legend that corresponds to the correlation plot colours? In my example 1 = red, 0 = white, and -1 = blue for the correlation plot.

EDIT[Still trying to solve this one, if anyone has any suggestions, it would be great]

You may have already tried this, but I would guess that the fill somehow needs to be mapped to a variable rather than set to a constant in the theme in your function, possibly in geom_rect(). That's where I might start if I was in your shoes, anyway! :) — aosmith, Jun 20 '20 at 17:50

How to add a second legend using ggpairs in R?

0 Answers0