VisuClust: VisuClust

In VisuClust: Visualisation of Clusters in Multivariate Data

Description

The VisuClust package provides functions to display the result of clusterings, if they are either disjoint (every observation belongs to one cluster) or fuzzy (for each observation i and cluster k there is a value u[i,k](i=1,...,nPoints; k=1,...,nClusters)), which states the membership of the i.th observation for the cluster k, where sum(u[i,k],k=1,...,nClusters) = 1). A disjoint clustering can be created with known hierarchical methods, if the desired number of clusters is provided, or by using k-means-clustering (for example hclust, kmeans from the package cluster). A fuzzy clustering can be created for example with fanny from the package cluster.

The basis of the visualization is Sammon's method Nonlinear Mapping (NLM), which displays all observations in a plane so that the distances in the multivariate space are approximated best (see sammon from the package MASS). The clusters in the NLM diagram can be indicated by coloring, drawing lines between similar point pairs, and other features.

Author(s)

Michael Sieger <[e-mail]>
Georg Ohmayer <[e-mail]>

References

Will be published in the GIL conference transcript 2012: Visualisierung von Clustern in multivariaten Daten unter Einsatz von R Georg Ohmayer, Michael Sieger

VisuClust documentation built on May 2, 2019, 6:53 a.m.