ebpmf_exponential_mixture: Empirical Bayes Poisson Matrix Factorization

Description Usage Arguments Details Value Examples

View source: R/ebpmf_exponential_mixture.R

Description

Uses Empirical Bayes to fit the model

X_{ij} ~ Poi(∑_k L_{ik} F_{jk})

with

L_{.k} ~ g_k()

with g_k being either Mixture of Exponential, or Point Gamma

Usage

1
2
ebpmf_exponential_mixture(X, K, m = 2, maxiter.out = 10,
  maxiter.int = 1, seed = 123)

Arguments

X

count matrix (dim(X) = c(n, p)).

K

number of topics

m

multiplicative parameter for selecting grid in "ebpm::ebpm_exponential_mixture"

maxiter.out

maximum iterations in the outer loop

maxiter.int

maximum iterations in the inner loop

seed

random seed

Details

The model is fit in 2 stages: i) estimate g by maximum likelihood (over pi_k) ii) Compute posterior distributions for λ_j given x_j,\hat{g}.

Value

A list containing elements:

qls_mean

A n by k matrix: Approximate posterior mean for L

qls_mean_log

A n by k matrix: Approximate posterior log mean for L

gls

A list of K elements, each element is the estimated prior for the kth column of L

qfs_mean

A p by k matrix: Approximate posterior mean for F

qfs_mean_log

A p by k matrix: Approximate posterior log mean for F

gfs

A list of K elements, each element is the estimated prior for the kth column of F

Examples

1
To add

zihao12/ebpmf documentation built on Nov. 5, 2019, 1:27 p.m.