Subsetting Data Frames

Data frames support three main subsetting mechanisms inherited from matrices and lists. The single-bracket operator with row and column indices (df[rows, cols]) selects a rectangular subset and always returns a data frame unless drop = TRUE or only one column is selected. The dollar-sign operator (df$column) extracts a single column as a vector and is the most common way to access or modify individual columns. The double-bracket operator (df[["column"]]) is equivalent to the dollar sign but allows programmatic column names stored in variables. Logical subsetting is the backbone of data filtering. A logical vector of the same length as the number of rows, passed as the row argument, selects only the matching rows. You build this vector using comparison and logical operators: ==, !=, <, >, <=, >=, &, |, and !. The subset() function provides a cleaner syntax for the same operation and additionally allows selecting columns by name without quotes, at the cost of not being safe inside functions due to non-standard evaluation. The which() function is useful when you need indices rather than a logical mask. Row selection can also be done with a character vector of row names if the data frame has meaningful row names set via rownames(df) <- .... Combining row and column selection in a single bracket expression gives you surgical control over any rectangular region of the data frame.

Example

df <- data.frame(
    name   = c("Alice", "Bob", "Carol", "Dave", "Eve"),
    age    = c(28, 34, 22, 41, 30),
    salary = c(55000, 72000, 48000, 95000, 61000),
    dept   = c("HR", "IT", "IT", "Finance", "HR")
)

# Dollar-sign: extract a column as a vector
df$name       # character vector of names
df$salary     # numeric vector of salaries

# Double-bracket: programmatic column access
col <- "age"
df[[col]]     # same as df$age

# Single-bracket: row/column matrix-style
df[1, ]           # first row (a 1-row data frame)
df[ , 2]          # second column (drops to vector by default)
df[ , 2, drop = FALSE]  # keep as data frame
df[1:3, c("name", "age")]   # rows 1-3, named columns

# Logical row filtering
df[df$age > 30, ]                      # rows where age > 30
df[df$dept == "IT", ]                  # IT employees
df[df$salary >= 60000 & df$active != FALSE, ]   # compound condition

# subset() — readable syntax (avoid inside functions)
subset(df, age < 35, select = c(name, salary))

# Negative indexing (drop columns by position)
df[ , -1]          # all columns except the first
df[ , -c(1, 4)]    # drop columns 1 and 4

# which() to get row indices
high_earners <- which(df$salary > 60000)
df[high_earners, ]

# Updating values in place via subsetting
df$salary[df$dept == "IT"] <- df$salary[df$dept == "IT"] * 1.05

HTML

Full Editor

Related Resources

R Reference

Complete tag & property list

R How-To Guides

Step-by-step practical guides

R Exercises

Practice what you've learned