rshudde/RJcluster: RJ Clustering Algorithm

Clustering algorithm for high dimensional data. This algorithm is ideal for data where N = P or N < P. Assuming that P feature measurements on N objects are arranged in an N×P matrix X, this package provides clustering based on the left Gram matrix XX^T. When the P-dimensional feature vectors of objects are drawn independently from a K distinct mixture distribution, the N-dimensional rows of the modified Gram matrix XX^T/P converges almost surely to K distinct cluster means. This transformation/projection thus allows the clusters to be tighter with order of P. To simulate data, type "help('simulate_HD_data')" and to learn how to use the clustering algorithm, type "help('RJclust')". To cite this package, type 'citation("RJcluster")'.

install.packages("remotes")
remotes::install_github("rshudde/RJcluster")