gibbs_basic: helper: make basic gibbs sampler code (incomplete)

In alexpkeil1/wellwise: WELL Water, Infants, and a Safe Environment: simulations

Description

not used by user, typically

Usage

gibbs_basic(x = c("x", "z"), z = c("bmi"), y = "y", binvars = NULL,
  matx = "X", standardizex = TRUE, binary = TRUE,
  xintv = rbind(c(0.99, 0), c(0, 0.99), c(0.99, 0.99)))

Arguments

x	intervenable exposures (character vector)
z	covariates (character vector)
y	outcome
binvars	= non-outcome variables that are binary (character vector)
matx	optional, name of matrix with intervenable exposures
standardizex	logical, should x be standardized?
binary	logical, is outcome binary?
xintv	matrix with ncol = number of intervenable exposures, nrow = number of interventions. Each value is on [0,1] and represents the proportional decrease in the value of x upon hypothetical intervention

Details

lorem ipsum

Examples

# library(rjags)
 dgm <- function(N=100, trueRD=0.2){
   x1 = rbinom(N, 1, 0.5)
   py00 = runif(N)*0.1 + 0.4
   l2 = rbinom(N, 1, 1/(1+exp(-1 + x1 + py00)))
   x2 = rbinom(N, 1, 1/(1+exp(-1 + x1 + l2)))
   py = py00 + trueRD*((x1 + x2)/2) #true risk difference per unit exposure;
   y = rbinom(N, 1, py)
   data.frame(x1, l2, x2, y)
 }
 dat = as.list(dgm(100))
 dat$N = 100
 dat$p = 5
 
 source("~/Epiprojects/wellwater/sims/code/make_stan_terms.R")
 
 mod = jags_basic(x=c('x1', 'x2'), z = 'l2', y='y', 
   binvars=c('x1', 'x2', 'l2'), xintv = rbind(c(1,0), c(0,1),c(1,1)), 
   binary=TRUE, matx = NULL)
 cat(mod)
# usage in jags (or edit by hand)
# not run
# tf = tempfile()
# cat(mod, file=tf)
# jags.model(file = tf, data = dat, n.chains=1)

alexpkeil1/wellwise documentation built on June 15, 2019, 7:02 p.m.