HOPACHConfig

Properties

dist

string

Distance measure.

one of"cosangle""abscosangle""euclid""cor""abscor"

level_selection

string

Which level of the tree becomes the returned partition: "best" = the level minimizing MSS, "greedy" = the first level below which MSS increases.

one of"best""greedy"

max_levels

integer≥ 1≤ 15

Maximum number of levels in the tree. Capped at 15, above which the level encoding overflows.

max_children

integer≥ 2≤ 9

Maximum number of children at each node.

max_children_mss

integer≥ 2≤ 9

Maximum number of children at each node when computing MSS.

collapse

string

How collapsing is performed at each level: "seq" = collapse pairs sequentially from the closest while MSS decreases, "all" = collapse any pair that decreases MSS.

one of"seq""all"

new_medoid

string

How the medoid of a collapsed pair is chosen: "medsil" = maximizer of medoid-based silhouette, "center" = minimizer of average distance to the medoid, "nn" = size-weighted nearest neighbor of the two medoids' mean, "uwnn" = unweighted nearest neighbor.

one of"medsil""center""nn""uwnn"

mss

string

Split silhouette criterion: "med" = median split silhouette, "mean" = mean split silhouette.

one of"med""mean"

min_improvement

number≥ 0≤ 1

Relative improvement in MSS required to accept a collapse step.

initial_order

string

How clusters in the first level are ordered: "co" = maximize correlation ordering, "clust" = order by a binary-split HOPACH of the cluster medoids.

one of"co""clust"

element_order

string

How elements are ordered within a cluster: "own" = by distance from the cluster's own medoid, "neighbor" = by distance to the next cluster's medoid.

one of"own""neighbor"

Relationships