plotting.R

library(dplyr)

# We will again be using the NPAS data set.

# Let's clean it up a bit
# (data from https://www.kaggle.com/lucasgreenwell/nerdy-personality-attributes-scale-responses)
df_NPAS <- read.csv("NPAS data.csv") |>
  # this is similar to na.omit() but for specific variables:
  tidyr::drop_na(Q1:Q26, gender, age) |>
  mutate(
    Nerdy = across(Q1:Q26, .fns = as.numeric) |> rowMeans(na.rm = TRUE),
    gender = factor(
      gender,
      levels = c(2, 1, 3),
      labels = c("woman", "man", "other")
    )
  ) |>
  select(-(Q1:Q26)) |>
  filter(age < 100) |>
  sample_n(200) # get just some of the data


glimpse(df_NPAS)


# *** ggplot2 *** ---------------------------------------------------------

library(ggplot2)

# When using `ggplot` there are two main concepts to understand:
#    1. Plots are built ("drawn") - layer by layer.
#    2. Data (variables) are mapped to visual features of the plot.
#     Like saying "on the x axis we have 'age'",
#     or "color represents 'gender'".
#
# This course will not expand much on using `ggplot`, as its use changes
# dramatically between users, data type and whole fields, but let's look at some
# of the basics.

# Basic steps - think:
# 1. What variables do want to plot?
# 2. What data do we want to plot? Raw data (individual scores)? Summarized data
#   (group means)?
# 2. What type of plot are we making? (a scatter plot? A bar plot? error bars?)
# 3. How do variables MAP onto the plot - in other words what visual feature
#   represents different variables? Does color change by group? Does location on
#   the X axis change according to age? Etc...

# Example 1 ---------------------------------------------------------------

?ggplot

# The main function is `ggplot()`, and it takes a data frame.
ggplot(df_NPAS)
# This drew nothing, because we did not tell it what to draw or how...

# `aes()` is the mapping function - it lets us MAP variables onto visual
# features. For example, I want the X-axis to represent `Nerdy`, and color to
# represent `gender`.
ggplot(df_NPAS, mapping = aes(x = Nerdy, color = gender))
# This still does nothing, because we didn't add any layers - we didn't tell
# ggplot what to draw!

# All the actual "drawing" is specified with the "geoms". Here we might want to
# draw a histogram. So lets add that - literally, with a `+`!
ggplot(df_NPAS, mapping = aes(x = Nerdy, color = gender)) +
  geom_histogram()
# Starting to get somewhere!

# Seems like we don't want the color to represent gender, but maybe the filling
# instead? Just MAP onto `fill`.
ggplot(df_NPAS, mapping = aes(x = Nerdy, fill = gender)) +
  geom_histogram()


# Or maybe I want a density plot instead?
ggplot(df_NPAS, mapping = aes(x = Nerdy, color = gender)) +
  geom_density()


ggplot(df_NPAS, mapping = aes(x = Nerdy, color = gender)) +
  geom_density(
    mapping = aes(linetype = gender),
    size = 1,
    fill = "yellow",
    alpha = 0.4
  )

# note that `linetype` is in `aes()` - so it varies according to some variable,
# whereas shape, fill, and alpha (the opacity) are not inside `aes()` so there
# are FIXED to a constant.

# ggplot(df_NPAS, mapping = aes(x = Nerdy, color = gender, linetype = gender)) +
#   geom_density(
#     size = 1,
#     fill = "yellow",
#     alpha = 0.4
#   )

# Example 2 ---------------------------------------------------------------

# Let's try another example, with the same data and variables (Nerdy & gender):
ggplot(df_NPAS, aes(x = gender, y = Nerdy))


ggplot(df_NPAS, aes(x = gender, y = Nerdy)) +
  geom_point()

# because the x-variable is categorical, we might get a better understanding of
# the data with a box plot:
ggplot(df_NPAS, aes(x = gender, y = Nerdy)) +
  geom_boxplot()


ggplot(df_NPAS, aes(x = gender, y = Nerdy)) +
  geom_boxplot() +
  geom_violin()


# Can't really see the boxes! Order of geoms matters!
ggplot(df_NPAS, aes(x = gender, y = Nerdy)) +
  geom_violin() +
  geom_boxplot(mapping = aes(color = gender), fill = NA)


# Note that both geom_violin and geom_boxplot didn't draw the raw data as is,
# but summarized the data first. In some cases, we might want to summarize the
# data ourselves in some way before plotting (with `group_by() |> summarize()`)

# Example 3 ---------------------------------------------------------------

# What will this draw?
ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(mapping = aes(color = gender))

# We can also map color on to a continuous variable:
ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = TIPI2))


ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = gender)) +
  geom_smooth() # add regression line


ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = gender)) +
  geom_smooth(aes(color = gender), method = "lm")


# we can split into subplots using facets:
ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = gender)) +
  geom_smooth(aes(color = gender), method = "lm") +
  facet_grid(rows = vars(urban))


# We can "prettify" the plot with themes, change the "scales", and more...
ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = gender), alpha = 0.7, shape = 3) +
  geom_smooth(aes(color = gender), method = "lm", size = 1.5, fill = "gray") +
  facet_grid(
    cols = vars(urban),
    # We can change the facet labels:
    labeller = as_labeller(c(
      "0" = "N/A",
      "1" = "Rural",
      "2" = "Suburban",
      "3" = "Urban"
    ))
  ) +
  # scale_*() functions can be used to control the appearance of different
  # scales (x, y, color, fill, size...) - things that we've mapped.
  scale_color_manual(
    values = c(woman = "red4", man = "steelblue4", other = "purple1"),
    labels = c("Woman", "Man", "Other")
  ) +
  coord_cartesian(ylim = c(0, 5)) +
  labs(x = "Age [years]", color = "Gender") +
  theme_light() +
  theme(legend.position = "bottom")
# Learn more about themes here:
# https://tidyverse.org/blog/2025/10/ggplot2-styling/

# and many many more...
# - The basics: https://ggplot2-book.org/
# - Explore options + cheat sheet here: https://ggplot2.tidyverse.org/
# - How to get the plot you want: https://www.r-graph-gallery.com/index.html
# - Learn how to better visualize your data: https://clauswilke.com/dataviz/
#
# ggplot is a powerful tool - with many other packages interfacing or expanding
# on it. We will see some of them later on.

# Take 5 minutes, look at the data - think what you'd like to plot.

# Hebrew / Arabic plots ---------------------------------------------------

# ggplot2 for the most part supports right-to-left (RTL) scripts:
# https://tidyverse.org/blog/2025/01/text-rendering-updates/

# Exporting plots ---------------------------------------------------------

# You "can" export images via the "Export" menu in the "Plots" tab,
# but this save the images in very poor quality.

# for best results, make sure {ragg} is installed:
library(ragg)

p <- ggplot(df_NPAS, aes(x = age, y = Nerdy)) +
  geom_point(aes(color = gender))

# As tiff
ggsave(
  filename = "p2.tiff",
  plot = p,
  # - Size -
  units = "mm",
  width = 480,
  height = 300,
  # - Resolution -
  dpi = 600,
  scaling = 1 # play with this one to get it juuuust right
)


# As png
ggsave(
  filename = "p2.png",
  plot = p,
  # - Size -
  units = "mm",
  width = 480,
  height = 300,
  # - Resolution -
  dpi = 600,
  scaling = 1 # play with this one to get it juuuust right
)

# As pdf
ggsave(
  filename = "p2.pdf",
  plot = p,
  # - Size -
  units = "mm",
  width = 480,
  height = 300,
  # - Resolution -
  dpi = 600,
  scale = 1 # play with this one to get it juuuust right
)