TGLG_binary_revised: TGLG model fitting for binary outcome

In y1zhong/TGLG: What the package does (short line)

Description

TGLG model fitting for binary outcome

Usage

TGLG_binary_revised(
  X,
  y,
  net = NULL,
  nsim = 30000,
  ntune = 10000,
  freqTune = 100,
  nest = 10000,
  nthin = 10,
  a.alpha = 0.01,
  b.alpha = 0.01,
  a.gamma = 0.01,
  b.gamma = 0.01,
  lambda.lower = 0,
  lambda.upper = 10,
  emu = -5,
  esd = 3,
  prob_select = 0.95
)

Arguments

a.alpha,	b.alpha: inverse-gamma distribution parameter for sigma_alpha
a.gamma,	b.gamma: inverse-gamma distribution parameter for sigma_gamma
lambda.lower,	lambda.uppper: lower and upper bound of uniform distribution for lambda
emu,	esd: mean and standard deviation of log-normal distribution for epsilon
X:	input features, dimension n*p, with n as sample size and p as number of features
y:	response variable: binary 0, 1
net:	igraph object that represents the network
nsim:	total number of MCMC iteration
ntune:	first number of MCMC iteration to adjust propose variance to get desired acceptance rate
freqTune:	frequency to tune the variance term.
nest:	number used for estimation
nthin:	thin MCMC chain

Value

post_summary: a dataframe including the selection probablity and posterior mean of betas

dat: X, y, and net used to generate results

save_mcmc: all the mcmc samples saved after burnin and thinning.

y1zhong/TGLG documentation built on July 18, 2021, 3:52 p.m.