tidyr expand_grid() in R: All Combinations From Vectors
The expand_grid() function in tidyr generates a tibble of all combinations of values from one or more vectors or lists. It is the tibble-friendly version of base R's expand.grid().
expand_grid(year = 2020:2024, product = c("X","Y","Z"))
expand_grid(x = 1:3, y = 1:2)
crossing(year = 2020:2024, product = c("X","Y")) # alias
base::expand.grid(...) # base R alternative
df |> expand(year, product) # different: from dataNeed explanation? Read on for examples and pitfalls.
What expand_grid() does in one sentence
expand_grid(...) returns a tibble where each row is a unique combination of values from the named arguments. Output rows = product of input lengths.
Syntax
expand_grid(...). ... is one or more named vectors / lists.
expand_grid() returns a TIBBLE; expand.grid() returns a data frame and reorders columns alphabetically. Use the tidyr version for cleaner integration with the tidyverse.Five common patterns
1. Two-vector grid
2. Three-vector grid
3. Custom values per column
4. Use for what-if analysis
5. Larger grids (caution: row count grows fast)
expand_grid() and crossing() are exact aliases. crossing was the original name; expand_grid was added later for clarity. Both work; pick one and stay consistent.expand_grid() vs crossing() vs expand() vs base expand.grid
| Function | Inputs | Output | Origin |
|---|---|---|---|
expand_grid(...) |
Named vectors | Tibble | tidyr |
crossing(...) |
Same | Tibble | tidyr (alias) |
expand(data, ...) |
Data frame + columns | Tibble | tidyr |
expand.grid(...) |
Named vectors | Data frame | base R |
When to use which:
- expand_grid / crossing for vector inputs in tidyverse code.
- expand for combining existing column values.
- expand.grid only when avoiding tidyverse.
A practical workflow
Use expand_grid for parameter sweeps and what-if analysis.
Compute sample size for every combination of alpha, power, and effect.
Common pitfalls
Pitfall 1: row count explosion. Three vectors of length 100 = 1,000,000 rows. Always check sizes first.
Pitfall 2: confusing with cross_join. expand_grid takes VECTORS; cross_join takes DATA FRAMES. They serve different inputs.
expand_grid returns rows in a SPECIFIC ORDER: the LAST argument varies fastest. expand_grid(a = 1:2, b = 1:3) returns (1,1), (1,2), (1,3), (2,1), (2,2), (2,3). The last column changes most rapidly.Try it yourself
Try it: Generate all combinations of 3 levels and 4 sizes. Save to ex_grid.
Click to reveal solution
Explanation: 3 levels * 4 sizes = 12 combinations.
Related tidyr / base functions
After mastering expand_grid, look at:
crossing(): alias for expand_gridexpand(): from existing datacomplete(): expand + merge with datanesting(): preserve column pairsdplyr::cross_join(): cartesian product of tablesbase::expand.grid(): base R alternative
FAQ
What does expand_grid do in tidyr?
expand_grid(...) returns a tibble of all combinations of values from named vector arguments. Each row is a unique combination.
What is the difference between expand_grid and expand.grid?
expand_grid (tidyr) returns a tibble and preserves argument order. expand.grid (base R) returns a data frame and reorders columns alphabetically.
What is the difference between expand_grid and crossing?
They are exact aliases. expand_grid is the newer name; crossing is the original. Both produce identical output.
What is the order of rows in expand_grid?
The LAST argument varies fastest. expand_grid(a=1:2, b=1:3) returns (1,1), (1,2), (1,3), (2,1), (2,2), (2,3).
Can expand_grid handle list inputs?
Yes. expand_grid(x = 1:3, y = list(c(1,2), c(3,4))) works for list inputs (each element is a row in the output).