How to sample random rows from a data frame in R

Sampling random rows from a data frame is essential for creating train/test splits, bootstrap resampling, and randomizing data order.

With dplyr

The tidyverse way uses slice_sample():

library(dplyr)

# Sample 5 random rows without replacement
sampled <- df %>% slice_sample(n = 5)

# Sample 10% of rows
sampled <- df %>% slice_sample(prop = 0.1)

# Sample with replacement (allows duplicate rows)
sampled <- df %>% slice_sample(n = 5, replace = TRUE)

# Weighted sampling (probability proportional to a column)
sampled <- df %>% slice_sample(n = 5, weight_by = some_column)

# Sample all rows in random order (shuffle)
shuffled <- df %>% slice_sample(prop = 1)

With base R

Base R uses sample() with row indices:

# Sample 5 random row indices without replacement
set.seed(42)  # for reproducibility
indices <- sample(nrow(df), size = 5)
sampled <- df[indices, ]

# Sample 10% of rows
n_to_sample <- round(nrow(df) * 0.1)
sampled <- df[sample(nrow(df), n_to_sample), ]

# Sample with replacement
sampled <- df[sample(nrow(df), 5, replace = TRUE), ]

# Randomize row order
shuffled <- df[sample(nrow(df)), ]

With data.table

library(data.table)

dt <- as.data.table(df)

# Sample 5 random rows without replacement
sampled <- dt[sample(.N, 5)]

# Sample 10% of rows
sampled <- dt[sample(.N, round(.N * 0.1))]

# Sample with replacement
sampled <- dt[sample(.N, 5, replace = TRUE)]

# Shuffle all rows
shuffled <- dt[sample(.N)]

Train/Test Split Example

A common use case is creating training and testing sets:

library(dplyr)

set.seed(123)

# Split: 70% train, 30% test
n <- nrow(df)
train_indices <- sample(n, size = 0.7 * n)

train <- df %>% slice(train_indices)
test  <- df %>% slice(-train_indices)  # remaining rows

Or with base R:

set.seed(123)
indices <- sample(nrow(df))
train_size <- round(0.7 * nrow(df))

train <- df[indices[1:train_size], ]
test  <- df[indices[(train_size + 1):n], ]

Stratified Sampling

Sample proportionally from groups:

library(dplyr)

# Sample 5 rows from each group
stratified <- df %>%
  group_by(category) %>%
  slice_sample(n = 5)

Common Variations

Task	dplyr	data.table
Sample n rows	slice_sample(n = 5)	dt[sample(.N, 5)]
Sample proportion	slice_sample(prop = 0.1)	dt[sample(.N, .N * 0.1)]
Sample with replacement	slice_sample(n = 5, replace = TRUE)	dt[sample(.N, 5, replace = TRUE)]
Shuffle rows	slice_sample(prop = 1)	dt[sample(.N)]

See Also

• sample() — Base R sampling function
• dplyr::slice() — dplyr row selection functions
• head() — Select first rows