dplyr min_rank() in R: Rank With Ties Sharing Min Position

The min_rank() function in dplyr ranks values where tied entries share the LOWEST available rank, leaving gaps. It is the "1224" or "competition" ranking style used in sports and academic standings.

min_rank(c(10, 20, 20, 30))         # 1, 2, 2, 4 (gaps after ties)
min_rank(desc(x))                    # rank descending
df |> mutate(rank = min_rank(score))
df |> group_by(g) |> mutate(rank = min_rank(score))
dense_rank(c(10, 20, 20, 30))       # 1, 2, 2, 3 (no gaps)
row_number(c(10, 20, 20, 30))       # 1, 2, 3, 4 (always unique)

What min_rank() does in one sentence

min_rank(x) returns the rank of each element where TIED values share the LOWEST rank and subsequent ranks SKIP to leave gaps. For c(10, 20, 20, 30), the result is c(1, 2, 2, 4).

This is the "competition" or "standard" ranking style. If two athletes share 2nd place, the next is 4th, not 3rd.

Syntax

min_rank(x). Use desc(x) for descending. NAs stay as NA in the output.

Five common patterns

1. Standard ascending rank

Tied 80s share rank 3. The 90 is rank 5 (rank 4 is skipped).

2. Descending rank

desc() reverses the comparison.

3. Top n via filter

Returns 4 rows because the two 80s tie at rank 2, and rank 3 is filled by the 90; total 4 in the top-3 ranks.

4. Per-group rank

Within each team, ranks restart.

5. NA handling

NAs stay as NA in the rank vector.

min_rank() vs dense_rank() vs row_number() vs percent_rank()

Four ranking functions in dplyr, with different tie behaviors.

When to use which:

• min_rank for sports / academic standings.
• dense_rank to count distinct rank levels.
• row_number for unique IDs, no shared ranks.
• percent_rank / cume_dist for percentile reasoning.

A practical workflow

Use min_rank when "what is my position relative to competitors" is the question.

The top 10 (with ties expanding the result if there are tied scores).

For per-category rankings:

Top 5 per category.

Common pitfalls

Pitfall 1: confusing min_rank with row_number. min_rank(c(10, 20, 20)) returns c(1, 2, 2). row_number(c(10, 20, 20)) returns c(1, 2, 3). The former shares ties; the latter breaks them.

Pitfall 2: filter by rank can return more than n rows. filter(min_rank(...) <= 3) may return more than 3 rows if there are ties at rank 3. To get exactly 3, use slice_max or row_number with <= 3.

Try it yourself

Related dplyr functions

After mastering min_rank, look at:

• dense_rank(): no-gap ranking
• row_number(): unique sequential IDs
• percent_rank() / cume_dist(): percentile-style ranks
• ntile(): bin into n equal groups
• base::rank(): average-tie default
• slice_max() / slice_min(): top/bottom n directly

For "give me exactly n rows", slice_max / slice_min are cleaner than rank-then-filter.

FAQ

What does min_rank do in dplyr?

min_rank(x) ranks values where tied entries share the LOWEST rank, with subsequent ranks skipping to leave a gap. Example: c(10, 20, 20, 30) becomes c(1, 2, 2, 4).

What is the difference between min_rank and dense_rank?

min_rank leaves gaps after ties (1, 2, 2, 4). dense_rank does not leave gaps (1, 2, 2, 3). dense_rank is useful when counting distinct rank levels.

What is the difference between min_rank and row_number?

min_rank shares ranks among ties. row_number always produces unique sequential ranks; ties are broken by row order. Use min_rank for "leaderboard"; row_number for unique IDs.

How do I rank descending?

Wrap in desc(): min_rank(desc(x)). Highest value gets rank 1.

How does min_rank handle NA?

NAs stay as NA in the output. Filter NAs before ranking, or use min_rank with the understanding that NAs don't get a rank.