Worksheet 10 — Factors: Ordering and Cleaning Categorical Data

Reorder, rename, and lump penguin categories with forcats — for cleaner plots and models

Bill Perry

2026-07-05

Factors — Taking Control of Categorical Data

Recap from the ANOVA worksheet

Fit a one-way ANOVA on penguin body mass by species
Found all three species differ; Gentoo heaviest
Every plot showed species in alphabetical order — today we fix that

Today’s Objectives (a 2-day worksheet)

Inspect factors and their levels
Reorder levels for readable bar charts and boxplots
Rename, collapse, and lump levels
Set the reference group for an ANOVA and drop unused levels

How to use this worksheet

Work through the parts in order. Type the code into a new R script in Positron and run it line by line.

Blocks marked ▶ Run this should be executed as written.

Blocks marked ✏️ Your turn ask you to write, modify, or interpret.

Day 1 = Parts 1–6. Day 2 = Parts 7–12.

🔮 Predict before you run — and type, don’t paste

Before each ▶ Run this block, predict the order the plot will use or the output you’ll see. Then type the code yourself. With factors the whole game is the order of the levels — predicting it first is how you learn to control it, and typing builds the muscle memory for the fct_ family.

🧩 Chunk 1 — What a factor is (Day 1, after lecture Chunk 1)

Parts 1–3: load the penguins and inspect their factors.

Part 1 · Load libraries and data

▶ Run this at the top of your script:

# Load packages + data -----------------------------------
library(tidyverse)      # includes forcats
library(palmerpenguins)

# Swap this line to use a different dataset later --------
penguins_df <- penguins %>% drop_na(species, body_mass_g)

⚠️ Watch out! forcats loads with library(tidyverse) — you do NOT need a separate library(forcats).

Part 2 · Inspect the factors

▶ Run this:

class(penguins_df$species)     # is it a factor?
levels(penguins_df$species)    # the ordered levels
fct_count(penguins_df$species) # count per level

✏️ Your turn: Record what you found.

Is species a factor?  Y / N
Levels, in order:
Which species has the most rows?

✏️ Your turn: Run the same three lines on island. How many levels does it have, and what are they?

# Write your code here:

island levels:

Part 3 · See the default (alphabetical) order

🔮 Predict first: In what order will the three species bars appear if you give no ordering instruction?

▶ Run this:

# ggplot uses the factor's level order = alphabetical ---
penguins_df %>%
  ggplot(aes(x = species)) +
  geom_bar() +
  theme_minimal()

✏️ Your turn: Was your prediction right? Why is alphabetical rarely the order you actually want?

Order shown:
Why alphabetical is a poor default here:

🧩 Chunk 2 — Reorder for readable plots (Day 1, after lecture Chunk 2)

Parts 4–6: reorder bars and boxplots by frequency and by value.

Part 4 · Order bars by frequency with `fct_infreq()`

▶ Run this:

# Most common species first ------------------------------
penguins_df %>%
  mutate(species = fct_infreq(species)) %>%
  ggplot(aes(x = species)) +
  geom_bar() +
  theme_minimal()

✏️ Your turn: Add fct_rev() to flip it to least-common-first. Write the one changed line.

# Write your modified mutate() line here:

Part 5 · Order a boxplot by value with `fct_reorder()`

🔮 Predict first: We’ll order species by their median body mass. Which species ends up last (heaviest)?

▶ Run this:

# Order species by median body mass ----------------------
penguins_df %>%
  mutate(species = fct_reorder(species, body_mass_g, .fun = median)) %>%
  ggplot(aes(x = species, y = body_mass_g, fill = species)) +
  geom_boxplot(alpha = 0.6) +
  theme_minimal() +
  theme(legend.position = "none")

✏️ Your turn: What order did the species come out in? Does the plot read more clearly than the alphabetical one?

New order (lightest -> heaviest):
Clearer?  Y / N

✏️ Your turn: Change .fun = median to .fun = mean. Did the order change? Why might mean and median give the same order here?

Your answer:

Part 6 · An ordered ranking plot

▶ Run this:

# Mean mass per species, biggest bar on top --------------
penguins_df %>%
  group_by(species) %>%
  summarize(mean_mass = mean(body_mass_g)) %>%
  mutate(species = fct_reorder(species, mean_mass) %>% fct_rev()) %>%
  ggplot(aes(x = mean_mass, y = species)) +
  geom_col(fill = "steelblue") +
  labs(x = "Mean mass (g)", y = NULL) +
  theme_minimal()

✏️ Your turn: What does fct_rev() do here? Remove it and describe what changes.

Your answer:

🛑 End of Day 1. You can now inspect and reorder factors. Day 2 covers renaming, lumping, and models.

🧩 Chunk 3 — Rename, collapse, lump (Day 2, after lecture Chunk 3)

Parts 7–9: clean up level labels.

Part 7 · Rename levels with `fct_recode()`

▶ Run this:

# Give levels report-ready names -------------------------
penguins_df %>%
  mutate(species = fct_recode(species,
    "Adélie penguin"    = "Adelie",
    "Chinstrap penguin" = "Chinstrap",
    "Gentoo penguin"    = "Gentoo"
  )) %>%
  count(species)

✏️ Your turn: In fct_recode(), which name goes on the left — the new one or the old one?

Your answer:

Part 8 · Collapse and lump levels

▶ Run this (collapse species into size groups):

penguins_df %>%
  mutate(size_group = fct_collapse(species,
    "smaller" = c("Adelie", "Chinstrap"),
    "larger"  = "Gentoo"
  )) %>%
  count(size_group)

▶ Run this (lump islands, keep the top 2):

penguins %>%
  drop_na(island) %>%
  mutate(island = fct_lump_n(island, n = 2)) %>%
  count(island)

✏️ Your turn: After lumping, what is the third island called, and how many rows landed in it?

Lumped level name:
n in that level:

Part 9 · The `as.numeric()` trap

🔮 Predict first: years below is a factor holding “2010”, “2011”, “2012”. What will as.numeric(years) return?

▶ Run this:

years <- factor(c("2010", "2011", "2012"))
as.numeric(years)                 # what comes out?
as.numeric(as.character(years))   # and this?

✏️ Your turn: Explain the difference in one sentence.

as.numeric(years) gave:
as.numeric(as.character(years)) gave:
Why they differ:

💡 Key idea: a factor is stored as integer codes with a labels table. as.numeric() returns the codes — always go through as.character() first when a factor holds real numbers.

🧩 Chunk 4 — Factors in models & cleanup (Day 2, after lecture Chunk 4)

Parts 10–12: set the reference, drop ghosts, build the final figure.

Part 10 · Set the ANOVA reference with `fct_relevel()`

🔮 Predict first: If we make Gentoo the reference group instead of Adelie, will the overall ANOVA p-value change, or only the coefficients?

▶ Run this:

# Default reference = Adelie -----------------------------
coef(lm(body_mass_g ~ species, data = penguins_df))

# Make Gentoo the reference ------------------------------
penguins_relevel <- penguins_df %>%
  mutate(species = fct_relevel(species, "Gentoo"))

coef(lm(body_mass_g ~ species, data = penguins_relevel))

✏️ Your turn: Compare the two coefficient sets.

What the (Intercept) represents now:
Did the coefficients change?  Y / N
Would the overall F-test / p-value change?  Y / N

Part 11 · Drop unused levels

▶ Run this:

# Filter out Gentoo — does the level disappear? ----------
two_species <- penguins_df %>% filter(species != "Gentoo")
levels(two_species$species)          # Gentoo still listed?

two_species <- two_species %>% mutate(species = fct_drop(species))
levels(two_species$species)          # now?

✏️ Your turn: Why does filter() leave the level behind, and what problem would that “ghost” level cause in a boxplot?

Your answer:

Part 12 · Put it all together

▶ Run this:

# Reordered + relabeled in one pipeline ------------------
penguins_df %>%
  mutate(
    species = fct_recode(species, "Adélie" = "Adelie"),
    species = fct_reorder(species, body_mass_g, .fun = median)
  ) %>%
  ggplot(aes(x = species, y = body_mass_g, fill = species)) +
  geom_boxplot(alpha = 0.6, outlier.shape = NA) +
  geom_point(position = position_jitter(width = 0.15, seed = 42),
             alpha = 0.3, size = 1.4) +
  labs(x = NULL, y = "Body mass (g)") +
  theme_minimal() +
  theme(legend.position = "none")

# Save it -----------------------------------------------
ggsave("figures/penguin_mass_by_species_ordered.png",
       width = 5, height = 5, units = "in", dpi = 300)

✏️ Your turn: Which two fct_ functions are doing the work in this figure, and what does each one do?

Function 1:                 does:
Function 2:                 does:

Part 13 · Review and checkpoint

At this point you should be able to:

Explain what a factor is and why level order matters
Inspect levels with levels() and fct_count()
Reorder with fct_infreq(), fct_reorder(), fct_rev()
Rename with fct_recode(), merge with fct_collapse(), lump with fct_lump_n()
Set a model reference with fct_relevel() and drop ghosts with fct_drop()
Explain why as.numeric() on a factor is a trap

✏️ Your turn — before you move on: Run your whole script top to bottom. Does it run cleanly?

Ran cleanly?  Y / N
If not, what error appeared:

Part 14 · Going further

Optional — do this if you finish early.

Reorder within groups on a line plot

▶ Try this:

# fct_reorder2: order legend to match the line ends ------
penguins_df %>%
  group_by(species, year = factor(year)) %>%
  summarize(mean_mass = mean(body_mass_g), .groups = "drop") %>%
  mutate(species = fct_reorder2(species, year, mean_mass)) %>%
  ggplot(aes(x = year, y = mean_mass, color = species, group = species)) +
  geom_line(linewidth = 1) +
  theme_minimal()

✏️ Your turn: How does fct_reorder2() decide the legend order, and why is that nice for line plots?

Your answer:

Your own data

✏️ Your turn: If you swap penguins for your project data, which categorical column would you turn into an ordered factor, and what would you order it by?

Your answer:

Getting unstuck

could not find function "fct_reorder" → library(tidyverse) (forcats is inside it).
Reordering “didn’t work” → you must reorder inside mutate() and then plot the result; reordering a copy doesn’t change the original.
A blank gap on the axis → a ghost level from filter(); add fct_drop().
Weird numbers from a factor → you hit the as.numeric() trap; use as.numeric(as.character(x)).
fct_recode did nothing → check the OLD name spelling; unknown old names are silently ignored.
Cheat sheet — https://rstudio.github.io/cheatsheets/factors.pdf

💡 Key idea: factors are how you make categorical variables behave — in plots and in models. Reorder for the reader, relabel for the report, relevel for the baseline.

End of Worksheet 10. Next: Worksheet 11 — joins, combining your data with a second table.

Worksheet 10 — Factors: Ordering and Cleaning Categorical Data

Factors — Taking Control of Categorical Data

Recap from the ANOVA worksheet

Today’s Objectives (a 2-day worksheet)

How to use this worksheet

🧩 Chunk 1 — What a factor is (Day 1, after lecture Chunk 1)

Part 1 · Load libraries and data

Part 2 · Inspect the factors

Part 3 · See the default (alphabetical) order

🧩 Chunk 2 — Reorder for readable plots (Day 1, after lecture Chunk 2)

Part 4 · Order bars by frequency with `fct_infreq()`

Part 5 · Order a boxplot by value with `fct_reorder()`

Part 6 · An ordered ranking plot

🧩 Chunk 3 — Rename, collapse, lump (Day 2, after lecture Chunk 3)

Part 7 · Rename levels with `fct_recode()`

Part 8 · Collapse and lump levels

Part 9 · The `as.numeric()` trap

🧩 Chunk 4 — Factors in models & cleanup (Day 2, after lecture Chunk 4)

Part 10 · Set the ANOVA reference with `fct_relevel()`

Part 11 · Drop unused levels

Part 12 · Put it all together

Part 13 · Review and checkpoint

Part 14 · Going further

Reorder within groups on a line plot

Your own data

Getting unstuck

Worksheet 10 — Factors: Ordering and Cleaning Categorical Data

Factors — Taking Control of Categorical Data

Recap from the ANOVA worksheet

Today’s Objectives (a 2-day worksheet)

How to use this worksheet

🧩 Chunk 1 — What a factor is (Day 1, after lecture Chunk 1)

Part 1 · Load libraries and data

Part 2 · Inspect the factors

Part 3 · See the default (alphabetical) order

🧩 Chunk 2 — Reorder for readable plots (Day 1, after lecture Chunk 2)

Part 4 · Order bars by frequency with fct_infreq()

Part 5 · Order a boxplot by value with fct_reorder()

Part 6 · An ordered ranking plot

🧩 Chunk 3 — Rename, collapse, lump (Day 2, after lecture Chunk 3)

Part 7 · Rename levels with fct_recode()

Part 8 · Collapse and lump levels

Part 9 · The as.numeric() trap

🧩 Chunk 4 — Factors in models & cleanup (Day 2, after lecture Chunk 4)

Part 10 · Set the ANOVA reference with fct_relevel()

Part 11 · Drop unused levels

Part 12 · Put it all together

Part 13 · Review and checkpoint

Part 14 · Going further

Reorder within groups on a line plot

Your own data

Getting unstuck

Part 4 · Order bars by frequency with `fct_infreq()`

Part 5 · Order a boxplot by value with `fct_reorder()`

Part 7 · Rename levels with `fct_recode()`

Part 9 · The `as.numeric()` trap

Part 10 · Set the ANOVA reference with `fct_relevel()`