KODAMA

An unsupervised and semi-supervised learning algorithm to performs feature extraction from noisy and high-dimensional data

News

KODAMA facilitates identification of patterns representing underlying groups on all samples in a data set. This is an improved version of KODAMA algorithm for spatially-aware dimensionality reduction. A landmarks procudere has been implemented to adapt the algorithm to the analysis of data set with more than 10,000 entries.

The KODAMA package has been integrated with t-SNE and UMAP to convert the KODAMA's dissimilarity matrix in a low dimensional space.

Installation

The KODAMA is avialable on https://CRAN.R-project.org/package=KODAMA.

library(devtools)
install_github("tkcaccia/KODAMAextra")