RHGC: Hierarchical Graph-based Clustering

In XuegongLab/HGC: A fast hierarchical graph-based clustering method

Description

Hierarchical clustering on a given undirected graph.

Usage

HGC.dendrogram(G)

Arguments

G	an object which represents the adjacency matrix of the graph, where G[i,j] is the weight of the edge between node i and node j, and zero means no link. The supported data structures include matrix, dgCMatrix, graph, and igraph.

Details

The function runs a hierarchical clustering on the given graph. It is a recursive procedure of two steps, first, the node pair sampling ratio is used as the distance metric to search the nearest neighbor pairs. Then the neighbor pair are merged and the graph is updated. The whole procedure is accelerated using the nearest neighbor chain algorithm. The algorithm stops when there's only one node left in the updated graph.

Value

An object of class hclust defined by the hclust function in the stats package. It is a list containing the clustering tree information with the components:

merge	an n-1 by 2 matrix. It records the two nodes in each merging step.
height	a set of n-1 real values. It is the height of the non-leaf nodes in the tree.
order	a vector giving the permutation of the original observations suitable for plotting.
labels	labels for the objects being clustered. Same as the rownames of G in default.
call	the call which produced the result.
method	the cluster method that has been used.
dist.method	the distance used here.

More details about the components are in the hclust.

Examples

data(Pollen)
Pollen.PCs <- Pollen[["PCs"]]
G <- SNN.Construction(Pollen.PCs, 25, 0.15)
tree = HGC.dendrogram(G)

XuegongLab/HGC documentation built on Dec. 18, 2021, 7:23 p.m.