ProjectionBasedClustering: Projection Based Clustering

A clustering approach for every projection method based on the generalized U*-matrix visualization of a topographic map is made available here. The number of clusters and the cluster structure can be estimated by counting the valleys in a topographic map. If the number of clusters and the clustering method are chosen correctly, then the clusters will be well separated by mountains in the visualization. Most projection methods are wrappers for already available methods in R. However, the neighbor retrieval visualizer (NeRV) is based on C++ source code of the 'dredviz' software package and the Curvilinear Component Analysis (CCA) is translated from 'MATLAB' ('SOM Toolbox' 2.0) to R.

Version: 1.0.1
Imports: Rcpp, ggplot2, stats, graphics, vegan, deldir, geometry, GeneralizedUmatrix, RcppEigen
LinkingTo: Rcpp
Suggests: fastICA, tsne, FastKNN, MASS, pcaPP, spdep, methods, pracma, grid, mgcv
Published: 2017-07-24
Author: Michael Thrun [aut, cre], Florian Lerch [aut], Felix Pape [aut], Kristian Nybo [cph], Jarkko Venna [cph]
Maintainer: Michael Thrun <m.thrun at>
License: GPL-3
NeedsCompilation: yes
CRAN checks: ProjectionBasedClustering results


Reference manual: ProjectionBasedClustering.pdf
Package source: ProjectionBasedClustering_1.0.1.tar.gz
Windows binaries: r-devel:, r-release:, r-oldrel:
OS X El Capitan binaries: r-release: ProjectionBasedClustering_1.0.1.tgz
OS X Mavericks binaries: r-oldrel: ProjectionBasedClustering_1.0.1.tgz
Old sources: ProjectionBasedClustering archive


Please use the canonical form to link to this page.