fec_stan: Modelling of faecal egg count data (one-sample case)

Description Usage Arguments Details Value Author(s) See Also Examples

View source: R/fec_stan.R

Description

Models faecal egg counts data in a one-sample case with (zero-inflated) Poisson-gamma model formulation using the Stan modelling language. It is computationally several-fold faster compare to conventional MCMC techniques. For the installation instruction of Stan, please read: Stan Installation.

Usage

1
2
3
4
fec_stan(fec, rawCounts = FALSE, CF = 50, zeroInflation = TRUE, 
  muPrior, kappaPrior, phiPrior, nsamples = 4000, 
  nburnin = 2000, thinning = 1, nchain = 1, 
  ncore = 1, adaptdelta = 0.95, verbose = FALSE)

Arguments

fec

vector of faecal egg counts

rawCounts

logical. If true, preFEC and postFEC correspond to raw counts (as counted on equipment). Otherwise they correspond to calculated epgs (raw counts times correction factor). Defaults to FALSE.

CF

correction factor or vector of correction factors

zeroInflation

logical. If true, uses the model with zero-inflation. Otherwise uses the model without zero-inflation

muPrior

a list with hyper-prior information for the baseline mean parameter μ. The default prior is list(priorDist = "gamma",hyperpars=c(1,0.001)), i.e. a gamma distribution with shape 1 and rate 0.001, its 90% probability mass lies between 51 and 2996

kappaPrior

a list with hyper-prior information for the dispersion parameter κ. The default prior is list(priorDist = "gamma",hyperpars=c(1,0.7)), i.e. a gamma distribution with shape 1 and rate 0.7, its 90% probability mass lies between 0.1 and 4.3 with a median of 1

phiPrior

a list with hyper-prior information for zero-inflation parameter. The default prior is list(priorDist = "beta",hyperpars=c(1,1))

nsamples

a positive integer specifying how many iterations for each chain (including burn-in samples)

nburnin

number of burn-in samples

thinning

thinning parameter, a positive integer specifying the period for saving samples

nchain

a positive integer specifying the number of chains

ncore

number of cores to use when executing the chains in parallel

adaptdelta

the target acceptance rate, a value between 0 and 1

verbose

logical. If true, prints progress and debugging information

Details

The first time each non-default model is applied, it can take up to 20 seconds for stan to compile the model. Currently the function only support prior distributions with two parameters. For a complete list of supported priors and their parameterization, please consult the list of distributions in Stan.

Sometimes the function outputs informational message from Stan regarding the Metropolis proposal rejections, this is due to the sampler hitting the boundary of a parameter space. For some variables, the boundary point is not supported in the distribution. This is not a concern if there are only a few such warnings.

The default number of iterations is 2000, with 1000 burn-in iterations. Normally this is sufficient in Stan. If MCMC chains do not converge, one should tune the MCMC parameters until convergence is reached to ensure correct results.

Value

Prints out summary of meanEPG as the posterior mean egg count. The posterior summary contains the mean, standard deviation (sd), 2.5%, 25%, 50%, 75% and 97.5% percentiles, the 95% highest posterior density interval (HPDLow95 and HPDHigh95) and the posterior mode. NOTE: we recommend to use the 95% HPD interval and the mode for further statistical analysis.

The returned value is a list that consists of:

stan.samples

An object of S4 class stanfit representing the fitted results. For more information, please see the stanfit-class in rstan reference manual.

posterior.summary

A data frame that is the same as the printed posterior summary.

Author(s)

Craig Wang [email protected]

See Also

simData1s for simulating faecal egg count data with one sample

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
## Not run: 
## load the sample data as a vector
data(echinococcus)
fec<-echinococcus[[1]]

## apply zero-infation model to the data vector
model<-fec_stan(fec,rawCounts=FALSE,CF=50)
samples<-stan2mcmc(model$stan.samples)

## a demonstration
demo("fecm_stan", package = "eggCounts") 

## End(Not run)

eggCounts documentation built on Aug. 3, 2017, 5:03 p.m.