Appearance

@graphty/algorithms / index / HierarchicalResult

Interface: HierarchicalResult ​

Defined in: indexed/hierarchical.ts:26

The dendrogram of an agglomerative clustering. Clusters 0 .. nodeCount - 1 are the nodes themselves; merge k creates cluster nodeCount + k from left[k] and right[k].

Properties ​

distance ​

readonly distance: Float64Array

Defined in: indexed/hierarchical.ts:34

Linkage distance of every merge.

height ​

readonly height: Uint32Array

Defined in: indexed/hierarchical.ts:36

Height of every cluster: 0 for a node, one more than its taller child for a merge.

left ​

readonly left: Uint32Array

Defined in: indexed/hierarchical.ts:30

First cluster of every merge.

nodeCount ​

readonly nodeCount: number

Defined in: indexed/hierarchical.ts:28

The node count the clustering ran over.

readonly right: Uint32Array

Defined in: indexed/hierarchical.ts:32

Second cluster of every merge.

roots ​

readonly roots: Uint32Array

Defined in: indexed/hierarchical.ts:42

The clusters left unmerged, in the order the legacy function lists its forest's trees: one entry (the last merge) when every pair of nodes could be linked, one per disconnected part otherwise, and none for an empty snapshot.

Methods ​

cut() ​

cut(height): Uint32Array<ArrayBufferLike>[]

Defined in: indexed/hierarchical.ts:54

Cut every root at height: a cluster no taller than height is kept whole, a taller one is replaced by its two children. The clusters come in left-to-right order, root by root.

Parameters ​

height ​

number

The cut height

Returns ​

Uint32Array<ArrayBufferLike>[]

cutAt() ​

cutAt(by): LabelResult

Defined in: indexed/hierarchical.ts:61

The flat partition after the first merges: with clusters: k, the first nodeCount - k merges, so exactly k clusters remain; with distance: d, the merges in order up to the first whose distance exceeds d.

Parameters ​

by ​

{ clusters: number; } | { distance: number; }

The cluster count, from roots.length to nodeCount, or the largest merge distance to apply

Returns ​

LabelResult

Throws ​

RangeError (code E_BAD_OPTION) for a cluster count outside that range or a NaN distance

members() ​

members(cluster): Uint32Array

Defined in: indexed/hierarchical.ts:48

The node indices of a cluster, left subtree first -- the order the legacy function's member Sets iterate in.

Parameters ​

cluster ​

number

A cluster index

Returns ​

Uint32Array