dplyr relocate() in R: Reorder Columns Easily
The relocate() function in dplyr moves columns to a specific position without renaming or dropping them. Use .before or .after to anchor the move; tidyselect helpers like where() and starts_with() let you move groups of columns at once.
relocate(df, hp) # move hp to front relocate(df, hp, .before = mpg) # move hp before mpg relocate(df, hp, .after = cyl) # move hp after cyl relocate(df, hp, .after = last_col()) # move hp to end relocate(df, where(is.character)) # move all chr cols to front relocate(df, starts_with("m")) # move m-cols to front relocate(df, name, .before = everything()) # name column first
Need explanation? Read on for examples and pitfalls.
What relocate() does in one sentence
relocate() moves columns to a specific position without renaming, dropping, or transforming them. You select columns with bare names or tidyselect helpers, then anchor the move with .before or .after. Without an anchor, the selected columns move to the front.
Unlike select(df, hp, mpg, everything()) (which renames implicitly via order), relocate() is a pure positioning operation. The data and names are unchanged; only the column order shifts.
Syntax
relocate() takes columns to move plus an optional .before or .after anchor. Without an anchor, columns move to the front (position 1).
The full signature:
relocate(.data, ..., .before = NULL, .after = NULL).before and .after are mutually exclusive. Both accept a single column name or a tidyselect expression.
last_col() to refer to the rightmost column. relocate(df, hp, .after = last_col()) moves hp to the end. relocate(df, hp, .before = last_col()) puts hp second-to-last. Useful when you do not know the exact column count.Six common patterns
1. Move column to the front
Without an anchor, the named columns move to position 1, in the order specified.
2. Move with .before
In this case .before = mpg produces the same result as no anchor (because mpg is currently first), but the intent is clearer.
3. Move with .after
.after = cyl places hp directly after cyl. Other columns shift right.
4. Move to the end
last_col() references whichever column is last after the move. Combined with .after, this puts the named column at the very end.
5. Move a group of columns by predicate
Tidyselect helpers like where(), starts_with(), contains() work inside relocate. Multiple columns move as a group.
6. Identifier columns first idiom
.before = everything() puts the column at the start. This is a common idiom for "make this the identifier column".
relocate() and select() can both reorder, but they differ in dropping behavior. relocate() keeps every column, only moving the named ones. select(df, mpg, cyl, everything()) keeps mpg first, cyl second, then everything else, but if you forget everything() it DROPS the unlisted columns. Use relocate() when you only want to move; use select() when you also want to drop.relocate() vs select() vs base R
Both verbs can reorder; only select() drops columns. Base R requires manual index manipulation.
| Task | relocate | select | Base R |
|---|---|---|---|
| Move to front | relocate(df, hp) |
select(df, hp, everything()) |
df[, c("hp", setdiff(names(df), "hp"))] |
| Move after specific col | relocate(df, hp, .after = cyl) |
(manual: list all in order) | (manual indexing) |
| Move all numeric | relocate(df, where(is.numeric)) |
select(df, where(is.numeric), everything()) |
(loop and reconstruct) |
| Move + drop | (does not drop) | select(df, hp, mpg) |
df[, c("hp","mpg")] |
| Reverse order | relocate(df, last_col():1) |
select(df, last_col():1) |
df[, ncol(df):1] |
When to use which:
- Use
relocate()when you ONLY want to move columns. - Use
select()when you want to move AND drop. - Use base R when zero dependencies matter.
Common pitfalls
Pitfall 1: forgetting everything() in select-based reorders. select(df, hp, mpg) returns ONLY hp and mpg, dropping the rest. To reorder without dropping, use relocate(df, hp, mpg) or select(df, hp, mpg, everything()).
Pitfall 2: .before and .after are mutually exclusive. Passing both errors. Choose one; if you need to position relative to two anchors (e.g., "between A and B"), use .before = B and rely on the columns to settle correctly.
relocate() does not change column COUNT, only order. This sounds obvious but bites users who confuse it with select(). If your row of names is shorter after relocate, you used the wrong verb.Pitfall 3: relocating into the middle of a tidyselect group. relocate(df, x, .after = where(is.numeric)) is ambiguous if there are multiple numeric columns. dplyr places after the LAST matching column. To be specific, name the anchor column directly.
Try it yourself
Try it: In mtcars, move hp and wt columns immediately after mpg. Save the result to ex_mt and print the column names.
Click to reveal solution
Explanation: relocate(hp, wt, .after = mpg) moves both hp and wt to the position right after mpg, in the order specified. Other columns (cyl, disp, etc.) keep their relative order.
Related dplyr functions
After mastering relocate(), look at:
select()witheverything(): alternative reorder that can also droprename(): change column names without moving themlast_col(): tidyselect helper for the rightmost columneverything(): tidyselect helper that matches all remaining columns- Base R
df[, order]: manual reorder using a name vector
For sorting column names alphabetically, the idiom is select(df, sort(tidyselect::peek_vars())) or simpler: select(df, sort(names(df))).
FAQ
How do I move a column to the front in dplyr?
relocate(df, col_name) moves it to position 1. Or relocate(df, col_name, .before = everything()) if you want to be explicit. Both produce the same result.
How do I reorder columns without dropping any in R?
Use relocate() from dplyr. It only changes positions; never drops columns. select(df, ...) reorders too but drops any columns you do not name unless you add , everything().
What is the difference between relocate and select in dplyr?
relocate() is positioning only: it never drops columns. select() can rename, drop, AND reorder, but you must explicitly include everything() if you want to keep unmentioned columns. For a pure reorder, prefer relocate().
How do I move multiple columns at once with relocate?
List them comma-separated: relocate(df, hp, wt, .after = mpg). Both move together as a block, in the order you specified, to the position after mpg.
How do I move columns to the end of a data frame?
Use .after = last_col(): relocate(df, mpg, .after = last_col()) moves mpg to the rightmost position. last_col() refers to whichever column is currently at the end.