Merge clusters based on variable bin thresholds — cluster_merge

Assigns each cluster to a bin for each variable based on where the cluster's median falls relative to the provided thresholds. Clusters that fall in the same combination of bins across all variables are merged.

Usage

cluster_merge_bin(data, cluster, thresholds)

Arguments

data: matrix or data.frame. Rows are observations, columns are variables. Must contain columns whose names match those in thresholds.
cluster: character vector. Cluster assignment for each observation. Length must equal nrow(data).
thresholds: named list. Each element is a numeric vector of threshold values for the variable given by the element name. The thresholds define bin boundaries: a single threshold t creates two bins (<= t and > t), two thresholds create three bins, and so on.

Value

A named list with two elements and a "thresholds" attribute:

assign: a tibble::tibble() with one row per observation and columns orig (original cluster label) and merged (merged cluster label).
label: a tibble::tibble() with one row per unique original cluster and columns orig (original cluster label), level (concise bin-combination label formed by joining per-variable bin indices with "_") and descriptive (human-readable bin-combination description formed by joining per-variable range descriptions with "; ").

The "thresholds" attribute stores the input thresholds list so that downstream functions such as cluster_merge_unimodal() can re-use the bin boundaries without requiring the caller to pass them again.

Examples

set.seed(1)
mat <- matrix(
  c(rnorm(20, 2), rnorm(20, -2)),
  ncol = 2,
  dimnames = list(NULL, c("var1", "var2"))
)
cl <- rep(c("A", "B", "C", "D"), each = 5)
thresholds <- list(var1 = 0, var2 = 0)
cluster_merge_bin(mat, cl, thresholds)
#> $assign
#> # A tibble: 20 × 2
#>    orig  merged
#>    <chr> <chr> 
#>  1 A     2_1   
#>  2 A     2_1   
#>  3 A     2_1   
#>  4 A     2_1   
#>  5 A     2_1   
#>  6 B     2_1   
#>  7 B     2_1   
#>  8 B     2_1   
#>  9 B     2_1   
#> 10 B     2_1   
#> 11 C     2_1   
#> 12 C     2_1   
#> 13 C     2_1   
#> 14 C     2_1   
#> 15 C     2_1   
#> 16 D     2_1   
#> 17 D     2_1   
#> 18 D     2_1   
#> 19 D     2_1   
#> 20 D     2_1   
#> 
#> $label
#> # A tibble: 4 × 3
#>   orig  level descriptive        
#>   <chr> <chr> <chr>              
#> 1 A     2_1   var1 > 0; var2 <= 0
#> 2 B     2_1   var1 > 0; var2 <= 0
#> 3 C     2_1   var1 > 0; var2 <= 0
#> 4 D     2_1   var1 > 0; var2 <= 0
#> 
#> attr(,"thresholds")
#> attr(,"thresholds")$var1
#> [1] 0
#> 
#> attr(,"thresholds")$var2
#> [1] 0
#>