BIRCH clustering (Geron Ch 8, morie.fn hmbrch)

Phase 1 streams points into leaf clustering features (N, LS, SS), absorbing into the nearest CF whose radius would stay under threshold; phase 3 does centroid-linkage agglomerative on the CF centroids. Labels are 0-based.

Usage

morie_geron_birch(X, n_clusters = 3, threshold = 0.5, branching_factor = 50)

Arguments

X

Points (n, d).

n_clusters

Final count or NULL.

threshold

Maximum sub-cluster radius.

branching_factor

Leaf cap.

Value

List with labels, subcluster_centers, subcluster_labels, subcluster_sizes, n_subclusters, radii, n_leaves.

Examples

V <- c(1, 2, 3, 4, 5, 6, 7, 8)
morie_geron_birch(V)
#> $labels
#> [1] 0 0 0 0 1 1 2 2
#> 
#> $subcluster_centers
#>      [,1]
#> [1,]  1.5
#> [2,]  3.5
#> [3,]  5.5
#> [4,]  7.5
#> 
#> $subcluster_labels
#> [1] 0 0 1 2
#> 
#> $subcluster_sizes
#> [1] 2 2 2 2
#> 
#> $n_subclusters
#> [1] 4
#> 
#> $radii
#> [1] 0.5 0.5 0.5 0.5
#> 
#> $n_leaves
#> [1] 1
#> 
#> $estimate
#> [1] 4
#> 
#> $n
#> [1] 8
#> 
#> $method
#> [1] "BIRCH: CF-tree leaf summarisation then centroid-linkage global clustering"
#>