Article: Just enough ggplot

Getting started

library(tidyverse)
theme_set(theme_bw(base_size = 12))
set.seed(1)

This website — https://ggplot2.tidyverse.org/reference/ — is incredibly useful. We’ll focus on the most basic building blocks. Let’s build a couple of data frames to use:

scores <- 
  tibble(
    name = c("mike", "carol", "greg", "marcia", "peter", "jan", "bobby", "cindy", "alice"),
    school = c("south", "south", "south", "south", "north", "north", "north", "south", "south"),
    teacher = c("johnson", "johnson", "johnson", "johnson",  "smith", "smith", "smith", "perry", "perry"),
    math_score = c(4, 3, 2, 4, 3, 4, 5, 4, 5),
    reading_score = c(1, 5, 2, 4, 5, 4, 1, 5, 4)
  )

random_numbers <- 
    tibble(
        binom = rbinom(1000, 10, 0.5),
        pois = rpois(1000, 5)
    ) %>% 
    mutate(
        mix = rnorm(1000, binom + pois, 5),
        high_low = ifelse(mix > 10, "high", "low")
    )

To get the basics down, I’ll show an example using the scores dataset and then ask you to make a similar plot using the random_numbers dataset.

Plotting one variable

geom_histogram

If your variable is continuous, you’ll want to use geom_histogram:

scores %>% 
    ggplot(aes(x = math_score)) +
    geom_histogram(binwidth = 2)

?geom_histogram

Make a histogram with the random numbers and play around with geom_histogram options:

random_numbers %>% 
  ggplot(aes(x = binom)) +
  geom_histogram(bins = 20)

geom_bar

If your variable is discrete, you’ll want to use geom_bar:

scores %>% 
    ggplot(aes(x = teacher)) +
    geom_bar()

Your turn with geom_bar:

random_numbers %>% 
  ggplot(aes(x = high_low)) +
  geom_bar() +
  coord_flip()

Plotting two variables

geom_point

It’s most common to use a scatter plot with geom_point:

scores %>% 
    ggplot(aes(x = math_score, y = reading_score)) +
    geom_point()

You don’t have to use continuous variables!

scores %>% 
    ggplot(aes(x = math_score, y = name)) +
    geom_point()

Your turn:

geom_col

Making a column for each “point” with geom_col can also be useful:

scores %>% 
    ggplot(aes(x = name, y = reading_score)) +
    geom_col() +
    geom_point(color = "red")

Your turn:

geom_boxplot

geom_boxplot is useful for plotting a discrete continuous variable:

scores %>% 
    ggplot(aes(x = school, y = math_score)) +
    geom_boxplot()

Your turn:

geom_smooth

geom_smooth is a fancy geom but still just uses two variables:

scores %>% ggplot(aes(x = reading_score, y = math_score)) + geom_smooth(method = "lm", se = FALSE)  + geom_point()

## `geom_smooth()` using formula 'y ~ x'

You can have multiple geoms — in this case it’s common to point points and a smoother both on a plot.

Your turn:

Adding a third variable

There are a few ways:

# color 
scores %>% 
    ggplot(aes(x = reading_score, y = math_score, color = school)) +
    geom_point()

# shape
scores %>% 
    ggplot(aes(x = reading_score, y = math_score, shape = school)) +
    geom_point()

# with a facet
scores %>% 
    ggplot(aes(x = reading_score, y = math_score)) +
    geom_point() +
    facet_wrap(~ school)

Your turn:

Putting it all together

Here’s a crazy plot that uses many, many options that might be useful for reference:

scores %>% 
    ggplot(mapping = 
             aes(x = math_score, y = reading_score, color = school)) +
    geom_point(alpha = 0.4) +
    scale_x_continuous(limits = c(0, 10)) +
    scale_y_log10() +
    geom_smooth(method = "lm", se = FALSE) +
    facet_wrap(~ school, ncol = 1) +
    labs(
        x = "Math",
        y = "Reading",
        title = "Math and reading test scores by student",
        subtitle = "Just for 8 students"
    ) +
    theme_gray(base_size = 18)

## `geom_smooth()` using formula 'y ~ x'

# this is useful for a session: theme_set(theme_bw(base_size = 12))

Your turn to make a crazy plot:

# go for it