If you are working with a massive data set which contains one or more categorical variables, and you feel that making one single plot does not result in anything interpretable due to overplotting, then you will be happy to know that ggplot has a couple of useful functions that create a grid of smaller plots, each displaying a specific level of the categorical variable(s). These functions are facet_grid() (further developed here) and facet_wrap() (further developed here).

Let’s take the following example to see what these functions do. In this example, there are 2 categorical predictor variables (group and category) and 2 continuous response variables (values_x and values_y). In total, there are 3000 entries in the resulting variables.

# predictor variables
group <- c(rep("A", 500), rep("B", 500), rep("C", 500), rep("D", 500), rep("E", 500), rep("F", 500))
category <- as.factor(floor(runif(3000, min=1, max=4)))

# response variables
values_x <- c(rnorm(500, mean=80, sd=10), rnorm(500, mean=100, sd=15), rnorm(500, mean=120, sd=10), rnorm(500, mean=75, sd=20), rnorm(500, mean=95, sd=10), rnorm(500, mean=115, sd=15))
values_y <- c(rnorm(500, mean=115, sd=10), rnorm(500, mean=125, sd=5), rnorm(500, mean=105, sd=10), rnorm(500, mean=85, sd=20), rnorm(500, mean=95, sd=10), rnorm(500, mean=90, sd=15))

# dataframe
df <- data.frame(group, category, values_x, values_y)

Here is a single scatter plot that displays ALL the data points at once. As you will realize, it is hardly revealing any specific pattern. NB: we will store the code in the object baseplot for later reuse:

baseplot <- ggplot(df, aes(values_x, values_y)) +
  geom_point()
baseplot

facet_wrap()

In our example, the first categorical variable group has 6 levels called A, B, C, D, E and F, and the second categorical variable category has 3 levels, 1, 2 and 3. facet_wrap() filters the data corresponding to all the combinations groupxcategory and makes as many mini-plots as necessary:

baseplot +
  facet_wrap(category~group)

facet_grid()

facet_grid() creates an organized matrix of mini-plots where the two categorical variables are individually plotted either on the X- or the Y-axis. The resulting matrix shows thus the same plots as with facet_wrap(s)) but in a different order:

baseplot +
  facet_grid(category~group)

In the following pages of this section, you will learn how to make use of these functions, how to modify their look, the labels and so on. Keep in mind that there is no universal solution for all data sets and that you may have to try several plots or designs before you find what represents best your data.

---