Adding italicised r with correlation coefficient to a scatter plot chart in ggplot

Question

I'm trying to use the code below to generate a simple scatter plot with a correlation coefficient that would have italicised r placed on the plot.

data(mtcars)

# Load required libraries
require(ggplot2)               # To derive the graphs
require(ggthemes)              # To apply ggplot themes to the chart
require(scales)                # For pretty breaks

# Function to generate correlation coefficient for the charts
corr_eqn <- function(x,y, digits = 2) {
  corr_coef <- round(cor(x, y), digits = digits)
  corr_coef <- expression(paste(italic(r)," = ", corr_coef))
  return(corr_coef)
}

# Provide a scatter plot for income and health deprivation
ggplot(mtcars, aes(x = drat, y = wt)) +
  geom_point(shape = 19, size = 2, aes(colour = as.factor(cyl))) +
  geom_smooth(colour = "red", fill = "lightgreen", method = 'lm') +
  ggtitle("Example") +
  xlab("drat") +
  ylab("wt") +
  scale_colour_tableau("tableau10") +
  geom_text(x = 3, y = 3,
            label = corr_eqn(mtcars$drat,
                             mtcars$wt), parse = TRUE) +
  theme(legend.key = element_blank(),
        legend.background = element_rect(colour = 'black'),
        legend.position = "bottom",
        legend.title = element_blank(),
        plot.title = element_text(lineheight = .8, face = "bold", vjust = 1),
        axis.text.x = element_text(size = 11, vjust = 0.5,
                                   hjust = 1, colour = 'black'),
        axis.text.y = element_text(size = 11, colour = 'black'),
        axis.title = element_text(size = 10, face = 'bold'),
        axis.line = element_line(colour = "black"),
        plot.background = element_rect(colour = 'black', size = 1),
        panel.background = element_blank())

The code stops with the ? mark in console. Running the code with the lines:

#   geom_text(x = 3, y = 3,
#             label = corr_eqn(mtcars$drat, mtcars$wt), parse = TRUE) +

commented, generates the following chart:

I'm guessing that my function to generate equation of the format r = 0.7 does not work, how can I fix it?

Solution

As you suspected, you just need to tweak your function. You could have used substitute as seen in this answer, but you can also just use paste here.

corr_eqn <- function(x,y, digits = 2) {
  corr_coef <- round(cor(x, y), digits = digits)
  paste("italic(r) == ", corr_coef)
}

Note that if you'd added as.character to what your original function returned things would have parsed. However, the result would have been with corr_coef as a string instead of the actual correlation coefficient you wanted.

I should also add that geom_text can result in poor resolution if you don't put the labels and coordinates into a new data.frame.

labels = data.frame(x = 3, y = 3, label = corr_eqn(mtcars$drat, mtcars$wt))

Then use the data argument and aes for geom_text:

geom_text(data = labels, aes(x = x, y = y,
                        label = label), parse = TRUE)

See annotate with geom = "text" as another option that avoids the new data.frame.

Comments

[MYaseen208]

Have a look on this similar question.

[Konrad]

@MYaseen208 as a matter of fact, I was trying to write my function using discussion in the linked question but it doesn't work. Clearly, I did something wrong.

[Konrad]

thank you very much for the comprehensive answer. Would it be very difficult to adjust the fonts, for instance by changing size and the font face?

[aosmith]

@Konrad You can definitely change the font size and family for the label in geom_text using the size and family arguments when using parse = TRUE, but not the font face.

[yPennylane]

corr_eqn <- function(x,y, digits = 2) { + corr_coef <- round(cor(x, y), digits = digits) + paste("italic(r) == ", corr_coef) + } labels = data.frame(x = 3, y = 3, label = corr_eqn(mtcars$drat, mtcars$wt)) produces a df with only one row ` x y label 3 3 italic(r) == -0.71`

[PatrickT]

Just to emphasize the point made by aosmith, It's not just that the labels must be stored in a dataframe, but they must also be passed as such inside geom_text, or to put it differently, passing labels$label to geom_text is not good enough.