HSI
From: Bayesian Models for Astrophysical Data, Cambridge Univ. Press
(c) 2017, Joseph M. Hilbe, Rafael S. de Souza and Emille E. O. Ishida
you are kindly asked to include the complete citation if you used this material in a publication
​
# Data from code 7.6
rztp <- function(N, lambda){
p <- runif(N, dpois(0, lambda),1)
ztp <- qpois(p, lambda)
return(ztp)
}
​
nobs <- 1000
x1 <- runif(nobs,-0.5,2.5)
xb <- 0.75 + 1.5*x1
exb <- exp(xb)
poy <- rztp(nobs, exb)
pdata <- data.frame(poy, x1)
xc <- -3 + 4.5*x1
pi <- 1/(1+exp((xc)))
bern <- rbinom(nobs,size =1, prob=1-pi)
pdata$poy <- pdata$poy*bern
​
​
​
​
Code 7.9 Zero-altered negative binomial (ZANB) or NB hurdle model in R using JAGS
==================================================================
require(R2jags)
Xc <- model.matrix(~ 1 + x1, data = pdata)
Xb <- model.matrix(~ 1 + x1, data = pdata)
Kc <- ncol(Xc)
Kb <- ncol(Xb)
model.data <- list(
Y = pdata$poy,
Xc = Xc,
Xb = Xb,
Kc = Kc, # number of betas − count
Kb = Kb, # number of gammas − binary
N = nrow(pdata),
Zeros = rep(0, nrow(pdata)))
sink("NBH.txt")
cat("
model{
# Priors beta and gamma
for (i in 1:Kc) {beta[i] ~ dnorm(0, 0.0001)}
for (i in 1:Kb) {gamma[i] ~ dnorm(0, 0.0001)}
# Prior for alpha
alpha ~ dunif(0.001, 5)
# Likelihood using zero trick
C <- 10000
for (i in 1:N) {
Zeros[i] ~ dpois(-ll[i] + C)
LogTruncNB[i] <- 1/alpha * log(u[i]) +
Y[i] * log(1 - u[i]) + loggam(Y[i] + 1/alpha) -
loggam(1/alpha) - loggam(Y[i] + 1) -
log(1 - (1 + alpha * mu[i])^(-1/alpha))
z[i] <- step(Y[i] - 0.0001)
l1[i] <- (1 - z[i]) * log(1 - Pi[i])
l2[i] <- z[i] * (log(Pi[i]) + LogTruncNB[i])
ll[i] <- l1[i] + l2[i]
u[i] <- 1/(1 + alpha * mu[i])
log(mu[i]) <- inprod(beta[], Xc[i,])
logit(Pi[i]) <- inprod(gamma[], Xb[i,])
}
}", fill = TRUE)
sink()
inits <- function () {
list(beta = rnorm(Kc, 0, 0.1),
gamma = rnorm(Kb, 0, 0.1),
numS = rnorm(1, 0, 25),
denomS = rnorm(1, 0, 1))}
params <- c("beta", "gamma", "alpha")
ZANB <- jags(data = model.data,
inits = inits,
parameters = params,
model = "NBH.txt",
n.thin = 1,
n.chains = 3,
n.burnin = 4000,
n.iter = 6000)
print(ZANB, intervals=c(0.025, 0.975), digits=3)
==================================================================
Output on screen:
​
Inference for Bugs model at "NBH.txt", fit using jags,
3 chains, each with 6000 iterations (first 4000 discarded)
n.sims = 6000 iterations saved
mu.vect sd.vect 2.5% 97.5% Rhat n.eff
alpha 0.002 0.001 0.001 0.005 1.001 4000
beta[1] 0.733 0.031 0.665 0.787 1.020 150
beta[2] 1.508 0.016 1.481 1.542 1.019 150
gamma[1] -2.926 0.228 -3.386 -2.491 1.001 5000
gamma[2] 4.675 0.312 4.097 5.304 1.001 5700
deviance 20004176.033 3.180 20004171.871 20004183.917 1.000 1
For each parameter, n.eff is a crude measure of effective sample size,
and Rhat is the potential scale reduction factor (at convergence, Rhat=1).
DIC info (using the rule, pD = var(deviance)/2)
pD = 5.1 and DIC = 20004181.1
DIC is an estimate of expected predictive error (lower deviance is better).