PNMFARD: Automatic rank determination PNMF based on euclidean...

In richardbeare/pNMF: Projective non-negative matrix factorization

Description

Automatic rank determination PNMF based on euclidean distance.

Usage

PNMFARD(
  X,
  nmfMod,
  tol = 1e-05,
  maxIter = 5000,
  verboseN = FALSE,
  zerotol = 1e-10
)

Arguments

X	Input data matrix
nmfMod	NMF model from the NMF package
tol	tolerance for stopping criteria
maxIter	Maximum number of iterations
verbose	Print status messages

Details

Implementation of "Automatic Rank Determination in Projective Nonnegative Matrix Factorization." Zhirong Yang, Zhanxing Zhu, Erkki Oja. In the 9th International Conference on Latent Variable Analysis and Signal Separation (LVA 2010), pages 514-521, St. Malo, France, 2010 Derived from matlab code by Z. Yang, https://sites.google.com/site/zhirongyangcs/ardpnmf

sigma and wnorm are stored in the "extra" slot of the nmf object

Value

Fitted NMF model, as defined in NMF package. The "extra" slot contains sigma and wnorm

See Also

PNMFARD

Examples

library(NMF)
setNMFMethod("PNMFARD", pNMF::PNMFARD)
mkD <- function(NOISE=TRUE) {
  n <- 1000 # rows
  counts <- c(30, 10, 20, 10, 15, 15) # samples
  syntheticNMF(n=n, r=counts, offset = NULL, noise = NOISE,
               factors = FALSE, seed = 99)
}
k<-mkD()
estim.r2 <- nmf(k, 16, method="PNMFARD", nrun=1, seed="nndsvd")
#wnorm and sigma in the extra slot
estim.r2@extra

richardbeare/pNMF documentation built on June 30, 2020, 6:33 p.m.