ClusterNodeSelection: Subcluster significance test

In erhcv: Equi-Rank Hierarchical Clustering Validation

Description

Test the significance of a subcluster directly under a given cluster. If it is not significant, the subcluster is eliminated and its element are merged in the main cluster. Bootstrap samples of the underlying data needs to be provided, as well as the structure under consideration.

Usage

ClusterNodeSelection(cluster, testPos, alpha, data, BootData)

Arguments

cluster	the main cluster (of the form provided by the function hclust2tree)
testPos	the position of the subcluster to test, directly under the main cluster
alpha	the confidence level for the tests
data	the underlying data
BootData	the dataframe of bootstrap samples of Spearman rho, with columns named "(i,j)", where "i" and "j" are different leaves

Details

The hypothesis testing is made with the matrix of Spearman's rho for a given dataset, see \insertCitegaisser2010testingerhcv.

Value

The main cluster, with or without the node under test, wether the hypothesis can be rejected or not.

Author(s)

Simon-Pierre Gadoury

References

\insertRef

gaisser2010testingerhcv

erhcv documentation built on May 1, 2019, 6:28 p.m.