dplyr add_count() in R: Add Group Count Without Summarising

The add_count() function in dplyr adds a column with the per-group row count to every row, WITHOUT collapsing the data. It is mutate()-style; count() is summarise()-style.

df |> add_count(g)                            # adds n column per group
df |> add_count(g, sort = TRUE)               # sort desc by n
df |> add_count(g, name = "group_size")       # rename added column
df |> add_count(g, wt = price)                # weighted (sum of price)
df |> group_by(g) |> mutate(n = n())          # equivalent
df |> count(g)                                # collapses (different result!)
df |> add_count(g) |> filter(n > 5)           # use n in downstream filter

What add_count() does in one sentence

add_count(df, ...) is df |> group_by(...) |> mutate(n = n()) |> ungroup(). It adds a column n containing the per-group row count, WITHOUT collapsing rows.

This is the "mutate" cousin of count(). Use it when you need the count alongside the original data, not as a summary.

Syntax

add_count(x, ..., wt = NULL, sort = FALSE, name = NULL). ... are grouping columns.

Five common patterns

1. Add count column

2. Filter to common groups

A clean idiom for "filter rare categories".

3. Sort by group size

sort = TRUE arranges rows so the largest group appears first.

4. Custom column name

Useful when n would clash with an existing column.

5. Weighted count

total_weight = sum of wt per cyl group. Same idea as tally(wt = ...).

add_count() vs count() vs add_tally()

Three ways to handle group counts in dplyr.

When to use which:

• add_count(g) : most common; one-step add a count column.
• count(g) : when you only need the summary table.
• add_tally() : when the data is already grouped and you want to keep all rows.
• tally() : when already grouped and you want a summary.

A practical workflow

The classic "filter rare values" pattern uses add_count + filter.

This is the cleanest base pattern for cleaning categorical variables. Equivalent without add_count:

Same result, more code, easier to make a mistake.

Common pitfalls

Pitfall 1: column name clash. If your data already has an n column, add_count() will silently overwrite it. Use name = "..." to avoid this.

Pitfall 2: confusing add_count with count. count(df, g) returns ONE row per group; add_count(df, g) returns ALL rows with n added. Easy to type the wrong one.

Try it yourself

Related count functions

After mastering add_count, look at:

• count(): collapse-style sister of add_count
• add_tally(): same as add_count but for already-grouped data
• tally(): collapse-style; already-grouped data
• n(): per-group row count inside mutate / summarise
• n_distinct(): count of unique values
• cur_group_rows(): row indexes within current group

For more complex group-aware mutations, use group_by() |> mutate() directly.

FAQ

What is the difference between add_count and count in dplyr?

count(df, g) collapses to one row per group with column n. add_count(df, g) keeps ALL rows and adds n as a new column. Use add_count when you need the count alongside the original data.

How do I add a count column to a data frame without summarising?

Use add_count(df, group_col). It adds n with the per-group row count and returns all rows.

What is the difference between add_count and add_tally?

add_count(df, g) does the grouping for you. add_tally(df) requires the data to already be grouped via group_by(). Otherwise identical.

How do I filter to common categories with add_count?

df |> add_count(category) |> filter(n >= threshold) keeps only rows whose category has at least threshold members.

Can I weight add_count?

Yes: add_count(df, g, wt = price) sums price per group instead of counting rows.