Learn R Programming  /  Data Structures  /  Factor

Factor

A factor stores categorical data — a fixed set of possible categories, such as Low/Medium/High. Factors tell R that a variable is categorical, which matters for both memory efficiency and statistical modelling.

Creating a Factor

R Code
sizes <- factor(c("Medium", "Small", "Large", "Medium"))
sizes
levels(sizes)
nlevels(sizes)
Output
[1] Medium Small  Large  Medium
Levels: Large Medium Small
[1] "Large"  "Medium" "Small" 
[1] 3
💡 Tip
By default R sorts factor levels alphabetically, which is why Large comes before Medium and Small above, even though that is not the natural size order. Fix this with ordered factors, shown below.

Behind the Scenes: Integer Codes

A factor is actually stored as integers, with a separate table of labels attached. This makes factors far more memory-efficient than storing repeated text.

R Code
as.integer(sizes)   # the underlying integer codes
as.character(sizes) # convert back to plain text
Output
[1] 2 3 1 2
[1] "Medium" "Small"  "Large"  "Medium"

Ordered Factors

When categories have a natural order (Low < Medium < High), create an ordered factor so R understands the ranking, not just the labels.

R Code
sizes2 <- factor(
  c("Medium", "Small", "Large"),
  levels  = c("Small", "Medium", "Large"),
  ordered = TRUE
)
sizes2
sizes2[1] > sizes2[2]   # Medium > Small? now this works!
Output
[1] Medium Small  Large 
Levels: Small < Medium < Large
[1] TRUE

Counting Categories with table()

R Code
table(sizes)
Output
sizes
 Large Medium  Small 
     1      2      1

Renaming or Combining Levels

R Code
levels(sizes)[levels(sizes) == "Medium"] <- "Mid"
sizes
Output
[1] Mid   Small Large Mid  
Levels: Large Mid Small

Why Factors Matter for Modelling

🚀 Advanced Example
When a factor is used as a predictor in lm() or glm(), R automatically converts it into dummy (indicator) variables, using the first level as the baseline for comparison.
R Code
grp <- factor(c("control","control","treated","treated"))
response <- c(4.1, 4.3, 6.0, 6.4)

model <- lm(response ~ grp)
coef(model)
Output
(Intercept)   grptreated 
       4.20         2.10

Key Points

  • A factor represents a fixed, known set of categories — internally stored as integers plus labels.
  • Use ordered = TRUE when categories have a meaningful rank.
  • table() is the fastest way to count how many observations fall in each category.
  • Statistical modelling functions automatically treat factors as categorical predictors.