opnmfR_ranksel_splithalf: Rank selection based on split-half similarity measures...

In kaurao/opnmfR: Orthogonal Projective Non-negative Matrix Factorization

Description

Rank selection based on split-half similarity measures (cosine, inner-product, cosine correlation, and adjusted rand index)

Usage

opnmfR_ranksel_splithalf(
  X,
  rs,
  W0 = NULL,
  use.rcpp = TRUE,
  nrepeat = 1,
  similarity = "inner",
  splits = NA,
  plots = TRUE,
  seed = NA,
  rtrue = NA,
  ...
)

Arguments

X	A matrix, rows are the number of features and columns are number of samples
rs	A vector of ranks to test for selection, if rs=NULL then 1:nrow(X) is used (default NULL)
W0	A string or matrix for initialization (default NULL)
use.rcpp	A logical, use opnmfRcpp() (default TRUE) dimensional data (default TRUE)
nrepeat	A number, number of splits (default 1)
similarity	.... (deafault "inner")
splits	A list, provided when user wants to define split of sample. Indecies of columns for one half of the split and rest are tanken as second split half (default NA)
plots	A logical, create a dot plot displaying the similarity measures for ranks provided and indicating the highest value for each measure (default TRUE)
seed	the set.seed used to select the split-half (default NA)
rtrue	the true rank of the input data if known (default NA)

Value

A list containing the similarity measures for all methods, the time taken for selection, the seed used for split-half, the selected rank based on the highest of each similarity measure

Examples

 result <- opnmfR_ranksel_splithalf()

kaurao/opnmfR documentation built on March 12, 2021, 4:15 a.m.