Description Usage Arguments Details Value Note References Examples
Zero-inflated Poisson data are count data with an excess number
of zeros. The ZIP model involves the Poisson parameter lambda
and the
prevalence parameter pi
.
1 2 3 4 |
data |
Matrix, data frame or data set with positive integers, including zeros and of the minimal length 10. |
prior.lambda |
Numeric vector containing minimum and maximum of a uniform
distribution used as prior for the Poisson parameter |
prior.pi |
Numeric vector containing parameters of a beta distribution
describing prior knowledge about prevalence (proportion of contaminated
samples), e.g., |
simulation |
Not used any longer. |
chains |
Positive single numeric value, number of independent MCMC chains (default 3). |
burn |
Positive single numeric value, length of the burn-in period (default 4000). |
thin |
Positive single numeric value (default 1). The samples from every
k-th iteration will be used for inference, where k is the value of thin.
Setting |
update |
Positive single numeric value, length of update iterations for estimation (default 5000). |
max.time |
Maximum time for which the function is allowed to extend the chains. Acceptable units include 'seconds', 'minutes', 'hours', 'days', 'weeks' (default "15minutes") (see autorun.jags). |
plots |
Logical, if |
The ZIP model applies to count data and can be interpreted as a mixture
distribution with one component comprising the 'true' zeros and another component
of Poisson distributed values with density parameter lambda
. The prevalence
parameter pi
refers to the proportion of the second, true non-zero
component.
The Bayesian model for estimation prevalence and lambda parameter has
in rjags/JAGS (originally BRugs/Winbugs) syntax following form
1 2 3 4 5 6 7 8 9 10 11 12 13 14 |
The function rrisk.BayesZIP
returns an instance of the
bayesmodelClass
class containing following information:
|
Logical, whether the model has converged (assessed by the user). |
|
Data frame containing statitsics of the posterior distribution. |
|
Data frame giving the joint posterior probability distribution. |
|
Names of the parameters jointly estimated by the Bayes model. |
|
Model in rjags/JAGS syntax as a character string. |
|
Number of independent MCMC chains. |
|
Length of burn-in period. |
|
Length of update iterations for estimation. |
The convergence of the model should be checked using the diagnostic plots.
Bohning, D., Dietz, E., Schlattman, P., Mendonca, L. and Kirchner, U. (1999). The zero-inflated Poisson model and the decayed, missing and filled teeth index in dental epidemiology. Journal of the Royal Statistical Society, Series A 162:195-209.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 | #------------------------------------------
# Example of ZIP model
#------------------------------------------
# generate ZIP data
pi <- 0.01
n <- 200
lambda <- 3.5
zip.data <- rep(0,n)
zip.data[sample(1:n,n*pi,replace=FALSE)]<-rpois(n*pi,lambda=lambda)
# estimate using Bayes model for zero inflated data
resZIP <- rrisk.BayesZIP(data = zip.data,
prior.lambda = c(0,100),
prior.pi = c(1,1),
burn = 4000,
max.time = '40seconds',
update = 4000)
resZIP
# compare with naive results ignoring ZIP model
pi.crude <- sum(zip.data>0)/n
lambda.crude <- mean(zip.data)
print(pi.crude)
print(lambda.crude)
resZIP@results
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.