Factors with forcats

The forcats package provides tools for working with factors, which are R’s data structure for categorical data.

library(forcats)

Factors#

R represents categorical data with factors. A factor is an integer vector with a levels attribute that stores a set of mappings between integers and categorical values. When you view a factor, R displays not the integers but the levels associated with them.

For example, R will display c("a", "c", "b", "a") with levels c("a", "b", "c") but will store c(1, 3, 2, 1) where 1 = a, 2 = b, and 3 = c.

R will display:

[1] a c b a
Levels: a b c

R will store:

[1] 1 3 2 1
attr(,"levels")
[1] "a" "b" "c"

Create a factor with factor():

factor(x = character(), levels, labels = levels, exclude = NA, ordered = is.ordered(x), nmax = NA): Convert a vector to a factor. Also as_factor().
```
f <- factor(c("a", "c", "b", "a"), levels = c("a", "b", "c"))
```

Return its levels with levels():

levels(x): Return/set the levels of a factor.
```
levels(f)
levels(f) <- c("x", "y", "z")
```

Use unclass() to see its structure.

Inspect Factors#

fct_count(f, sort = FALSE, prop = FALSE): Count the number of values with each level.
```
fct_count(f)
```
fct_match(f, lvls): Check for lvls in f.
```
fct_match(f, "a")
```
fct_unique(f): Return the unique values, removing duplicates.
```
fct_unique(f)
```

Combine Factors#

fct_c(...): Combine factors with different levels. Also fct_cross().

f1 <- factor(c("a", "c"))
f2 <- factor(c("b", "a"))
fct_c(f1, f2)

fct_unify(fs, levels = lvls_union(fs)): Standardize levels across a list of factors.
```
fct_unify(list(f2, f1))
```

Change the order of levels#

fct_relevel(.f, ..., after = 0L): Manually reorder factor levels.
```
fct_relevel(f, c("b", "c", "a"))
```
fct_infreq(f, ordered = NA): Reorder levels by the frequency in which they appear in the data (highest frequency first). Also fct_inseq().
```
f3 <- factor(c("c", "c", "a"))
fct_infreq(f3)
```
fct_inorder(f, ordered = NA): Reorder levels by order in which they appear in the data.
```
fct_inorder(f2)
```

fct_rev(f): Reverse level order.

f4 <- factor(c("a", "b", "c"))
fct_rev(f4)

fct_shift(f): Shift levels to left or right, wrapping around end.
```
fct_shift(f4)
```
fct_shuffle(f, n = 1L): Randomly permute order of factor levels.
```
fct_shuffle(f4)
```
fct_reorder(.f, .x, .fun = median, ..., .desc = FALSE): Reorder levels by their relationship with another variable.
```
boxplot(PlantGrowth, weight ~ fct_reorder(group, weight))
```
fct_reorder2(.f, .x, .y, .fun = last2, ..., .desc = TRUE): Reorder levels by their final values when plotted with two other variables.
```
ggplot(
  diamonds,
  aes(carat, price, color = fct_reorder2(color, carat, price))
   ) + 
  geom_smooth()
```

Change the value of levels#

fct_recode(.f, ...): Manually change levels. Also fct_relabel() which obeys purrr::map syntax to apply a function or expression to each level.
```
fct_recode(f, v = "a", x = "b", z = "c")
fct_relabel(f, ~ paste0("x", .x))
```
fct_anon(f, prefix = ""): Anonymize levels with random integers.
```
fct_anon(f)
```
fct_collapse(.f, …, other_level = NULL): Collapse levels into manually defined groups.
```
fct_collapse(f, x = c("a", "b"))
```
fct_lump_min(f, min, w = NULL, other_level = "Other"): Lumps together factors that appear fewer than min times. Also fct_lump_n(), fct_lump_prop(), and fct_lump_lowfreq().
```
fct_lump_min(f, min = 2)
```
fct_other(f, keep, drop, other_level = "Other"): Replace levels with “other.”
```
fct_other(f, keep = c("a", "b"))
```

Add or drop levels#

fct_drop(f, only): Drop unused levels.

f5 <- factor(c("a", "b"), c("a", "b", "x"))
f6 <- fct_drop(f5)

fct_expand(f, ...): Add levels to a factor.
```
fct_expand(f6, "x")
```
fct_na_value_to_level(f, level = "(Missing)"): Assigns a level to NAs to ensure they appear in plots, etc.
```
f <- factor(c("a", "b", NA))
fct_na_value_to_level(f, level = "(Missing)")
```