backfitting: backfitting to correct K-factor model

In kkdey/flashr: Factor Loading Adaptive SHrinkage in R

Description

Correct the factor and loading matrix estimates using backfitting algorithm

Usage

backfitting(Y, initial_list, maxiter_bf = 100, flash_para = list(),
  gvalue = c("lik", "eigen"), parallel = FALSE)

Arguments

Y	is the data matrix (N by P)
initial_list	is the list from initial_list algorithm
maxiter_bf	maximum number of iterations
gvalue	is for output of the backfit, "eigen" means just provide the sum of square "lik" mean provide the lowerbound
Fest	is estimate for f to correct
flash_parais	the list for flash parameters setting up
initial_list	initial list for the starting value which contains l,f,l2,f2 priorpost_vec clik_vec

Details

Repeatedly applies rank 1 algorithm to Y-L[,-i]F[,-i]'

Value

list of factor, loading and variance of noise matrix

• l is a N by K matrix for loadings

• f is a P by K matrix for factors

• sigmae2 is variance for the error, scalar for constant case and vector for nonconstant case

Examples

N = 100
P = 200
Y = matrix(rnorm(N*P,0,1),ncol=P)
g = initial_list(Y,K = 10)
gb = backfitting(Y,g$l,g$f,maxiter_bf=100,maxiter_r1 = 5)

kkdey/flashr documentation built on May 20, 2019, 10:36 a.m.