CaliCo (version 0.1.1)

calibrate: Generates calibrate.class objects

Description

calibrate is a function that allows to generate a calibrate.class class in which the estimation is done for a defined model.class and prior.class objects.

Usage

calibrate(md, pr, opt.estim, opt.valid = NULL)

Arguments

md

a model.class object

pr

a prior.class object

opt.estim

estimation options

• Nmh Number of iteration of the Metropolis Hastings algorithm

• thetaInit Initial point

• r regulation percentage in the modification of the k in the Metropolis Hastings

• sig Covariance matrix for the proposition distribution (\(k*sig\))

• Nchains Number of MCMC chains to run (if Nchain>1 an output is created called mcmc which is a coda object codamenu)

• burnIn Number of iteration to withdraw

opt.valid

list of cross validation options (default value opt.valid=NULL)

• nCV Number of iterations for the cross validation

• type.valid Type of cross validation selected. "loo" (leave one out) is the only method emplemented so far.

Value

calibrate returns a calibrate.class object. Two main methods are available:

• print(mdfit) returns the main information concerning the calibrate.class object

See Also

prior, calibrate, forecast, sequentialDesign

Examples

# NOT RUN {
###################### The code to calibrate
X <- cbind(seq(0,1,length.out=10),seq(0,1,length.out=10))
code <- function(X,theta)
{
  return((6*X[,1]*theta[2]-2)^2*theta[1]*sin(theta[3]*X[,2]-4))
}
Yexp <- code(X,c(1,1,11))+rnorm(10,0,0.1)

############### For the first model
###### Definition of the model
md <- model(code,X,Yexp,"model1")
###### Definition of the prior densities
pr <- prior(type.prior=c("gaussian","gaussian","gaussian","gamma"),opt.prior=
list(c(1,0.01),c(1,0.01),c(11,3),c(2,0.1)))
###### Definition of the calibration options
opt.estim=list(Ngibbs=200,Nmh=400,thetaInit=c(1,1,11,0.1),r=c(0.3,0.3),
sig=diag(4),Nchains=1,burnIn=100)
###### Run the calibration
mdfit <- calibrate(md,pr,opt.estim)
####### The plot generated is a list of ggplot
p <- plot(mdfit,X[,1])
p$out
print(mdfit)
# }