zilgm_poisson: Neiborhood selection under the zero-inflated Poisson...

In bbeomjin/ZILGM: Implementations of local Markov random fields for count data with zero-inflation and overdispersion

Neiborhood selection under the zero-inflated Poisson distribution for Zero-inflated Local Graphical Model

Description

Neiborhood selection under the zero-inflated Poisson distribution I by zero-inflated Poisson regression with l_1-regularization.

Usage

zilgm_poisson(y, x, lambda, weights = NULL, update_type = c("IRLS", "MM"),
              penalty.factor = NULL, thresh = 1e-6, EM_tol = 1e-5, EM_iter = 3e+2,
 			  tol = 1e-6, maxit = 3e+2)

Arguments

y	A numeric vector. A response variable y.
x	A design matrix x.
lambda	A regularization parameter to control a level of l_1-penalty.
weights	Weights vector for observations. A default value for each observation is 1.
update_type	Types of algorithm for estimating coefficients.
penalty.factor	Weights vector for coefficients of each variable. A default value for each variable is 1.
thresh	Threshold value for the estimated coefficients.
EM_tol	Convergence tolerance for EM algorithm.
EM_iter	A integer value. Maximum number of EM algorithm iterations.
tol	Convergence tolerance for coordinate descent.
maxit	A integer value. Maximum number of coordinate descent iterations.

Value

An S3 object with the following slots

beta	Estimated coefficients vector.
prob	Estimated probability of structural zero.
pos_zero	Indices of zero values.
iteration	Iteration numbers until convergence.
loglik	l_1-penalized negative log-likelihood value.
z	Estimated latent variable.
call	The matched call.

References

Wang, Z., S. Ma, M. Zappitelli, C. Parikh, C.-Y. Wang, and P. Devarajan, 2016: Penalized count data regression with application to hospital stay after pediatric cardiac surgery. Stat. Methods Med. Res., 25, no.6, 2685-2703.
Choi, H., J. Gim, S. Won, Y. J. Kim, S. Kwon, and C. Park, 2017: Network analysis for count data with excess zeros. BMC genetics, 18, no. 1, 1-10.
Park, B., H. Choi, C. Park, 2021: Negative binomial graphical model with excess zeros.

Examples

# Zero-inflated Poissn regression with l_1 penalty
require(ZILGM)
set.seed(1)
n = 100; p = 10; prob = 2 / p;
A = generate_network(p, prob, type = "random")
simul_dat = zilgm_sim(A = A, n = n, p = p, zlvs = 0.1, 
					  family = "negbin", signal = 1.5, theta = 0.5, noise = 0.0)
y = simul_dat$X[, 1]
X = simul_dat$X[, -1]
poisson_fit = zilgm_poisson(y = y, x = X, lambda = 1, update_type = "IRLS")

bbeomjin/ZILGM documentation built on Aug. 5, 2023, 5:52 a.m.