The analysis of high-dimensional datasets often requires to extract meaningful variables from the data or compress the data into a more tractable number of features. In technical terms, for a high-dimensional dataset X with N samples and P dimensions (traits), dimensionality reduction techniques aim: i) to provide a meaningful low-dimensional representation Z of K dimensions while only losing minor amounts of information, ii) to use only a small number of free parameters, iii) to preserve the quantities of interest in the data. There are a variety of approaches for dimensionality reduction with different underlying mathematical concepts and parameters. This package provides a simple function to access 12 dimensionality reduction methods (DiffusionMap, DRR, ICA, LLE, Isomap, LaplacianEigenmap, MDS, PCA, kPCA, nMDS, tSNE and UMAP). It introduces a new stability criterion, which identifies the lower dimensional componet Z that can reliably be found in cross-validation on different subsets of M samples and P traits.
|License||MIT + file LICENSE|
|Package repository||View on GitHub|
Install the latest version of this package by entering the following in R:
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.