This R-package has Non-Negative Matrix Factorization (NNMF) functions re-programmed from the Mathematica package NonNegativeMatrixFactorization.m, [AAp1].

The implementation follows the description of the hybrid algorithm GD-CLS (Gradient Descent with Constrained Least Squares) in the article [FS1].

The functions is based on base R's sparse matrix library Matrix. (The reason I wrote this package is that there was no R library for NNMF using sparse matrices.)

To install use the command:

devtools::install_github("antononcube/R-packages", subdir = "NonNegativeMatrixFactorization")   

[AAp1] Anton Antonov, "Implementation of the Non-Negative Matrix Factorization algorithm in Mathematica", NonNegativeMatrixFactorization.m, (2013), MathematicaForPrediction at GitHub.

[FS1] Shahnaz, F., Berry, M., Pauca, V., Plemmons, R., "Document clustering using nonnegative matrix factorization", (2006), Information Processing & Management 42 (2), 373-386.