Skip to contents

Collapse duplicate key groups, concatenating selected columns and replacing other divergences with NA

collapse_by_keys.Rd

collapse_by_keys() enforces uniqueness of key combinations by collapsing (merging) rows that share the same keys (provided in ...). The output contains exactly one row per unique key combination.

For each non-key column within a key group:

  • If the column has a single distinct value within the group (optionally ignoring NA according to na_rm), that value is kept.

  • If the column has multiple distinct values:

    • If the column is selected by .concat, the distinct values are concatenated as a single character string using sep (after optional NA removal).

    • Otherwise, the column is set to NA for that collapsed row (typed NA via vctrs::vec_cast()).

When warn = TRUE, the function emits a warning listing the non-key columns (excluding .concat) that were divergent in at least one group and therefore were replaced by NA (for affected groups).

Usage

collapse_by_keys(
  .data,
  ...,
  .keys = NULL,
  .concat = NULL,
  sep = " ; ",
  na_rm = TRUE,
  warn = TRUE
)

Arguments

.data

A data frame or tibble.

...

Key columns defining groups that must be unique in the output. Uses tidyeval.

.keys

Optional alternative to …for programmatic key selection. Accepts either (i) a character vector of column names or (ii) a tidyselect expression evaluated in.data. If supplied, .keystakes precedence over…

.concat

Optional tidyselect specification of non-key columns whose divergent values should be concatenated. If NULL (default), no columns are concatenated.

sep

String used to separate concatenated values.

na_rm

Logical. If TRUE, ignore NA values when assessing distinctness within a group and when concatenating. If FALSE, NA participates in distinctness (i.e., c("A", NA) is divergent).

warn

Logical. If TRUE, emit a warning listing columns (excluding keys and .concat) that were divergent in at least one group and were replaced by NA.

Value

A tibble with one row per unique key combination.

Details

Type behavior

  • Columns selected in .concat return a character result when concatenation is needed. This may change the type of those columns in the output.

  • For non-.concat columns, divergent groups are replaced by a typed NA using vctrs::vec_cast(NA, x) to preserve the column's type whenever possible.

The output is an ungrouped tibble (.groups = "drop").

Examples

library(dplyr)
#> 
#> Attaching package: ‘dplyr’
#> The following objects are masked from ‘package:stats’:
#> 
#>     filter, lag
#> The following objects are masked from ‘package:base’:
#> 
#>     intersect, setdiff, setequal, union
library(tibble)

df <- tibble(
  exam.num_collec  = c(1, 1, 1, 2, 2),
  mat.matrice      = c("SER", "SER", "SER", "PLAS", "PLAS"),
  spe.denomination = c("E. coli", "E. coli", "E. coli", "S. aureus", "S. aureus"),
  result           = c("POS", "NEG", "POS", "NEG", "NEG"),
  value            = c(10, 10, 12, 5, 5),
  commentaire      = c("first", NA, "repeat", "ok", "ok"),
  source_info      = c("labA", "labA", "labB", "labC", NA),
  flag             = c(TRUE, TRUE, TRUE, FALSE, FALSE)
)

# Collapse by keys:
# - concatenate selected text columns if needed
# - replace other divergent columns by NA
out <- df %>%
  collapse_by_keys(
    exam.num_collec, mat.matrice, spe.denomination,
    .concat = c(commentaire, source_info),
    sep = " | ",
    na_rm = TRUE,
    warn = TRUE
  )
#> Warning: Divergent columns replaced by NA: result, value

out
#> # A tibble: 2 × 8
#>   exam.num_collec mat.matrice spe.denomination result value commentaire   
#>             <dbl> <chr>       <chr>            <chr>  <dbl> <chr>         
#> 1               1 SER         E. coli          NA        NA first | repeat
#> 2               2 PLAS        S. aureus        NEG        5 ok            
#> # ℹ 2 more variables: source_info <chr>, flag <lgl>