CGP (version 2.1-1)

summary.CGP: CGP model summary information

Description

Print a summary of a ``CGP'' object.

Usage

# S3 method for CGP
summary(object, ...)

Arguments

object

An object of class "CGP"

…

For compatibility with generic method summary

Value

This function prints the results of:

lambda

Estimated nugget value \((\lambda)\)

theta

Estimated correlation parameters \((\theta)\) in the global GP

alpha

Estimated correlation parameters \((\alpha)\) in the local GP

bandwidth

Estimated bandwidth parameter \((b)\) in the variance model

rmscv

Root mean squared (leave-one-out) cross validation error

mu

Estimated mean value \((\mu)\) for global trend

tau2

Estimated variance \((\tau^2)\) for global trend

beststart

Best starting value found for optimizing the log-likelihood

objval

Optimal objective value for the negative log-likelihood (up to a constant)

Details

This function prints a summary of a ``CGP'' object.

References

Ba, S. and V. Roshan Joseph (2012) ``Composite Gaussian Process Models for Emulating Expensive Functions''. Annals of Applied Statistics, 6, 1838-1860.

See Also

CGP, print.CGP, predict.CGP

Examples

# NOT RUN {
x1<-c(0,.02,.075,.08,.14,.15,.155,.156,.18,.22,.29,.32,.36,
      .37,.42,.5,.57,.63,.72,.785,.8,.84,.925,1)
x2<-c(.29,.02,.12,.58,.38,.87,.01,.12,.22,.08,.34,.185,.64,
      .02,.93,.15,.42,.71,1,0,.21,.5,.785,.21)
X<-cbind(x1,x2)
yobs<-sin(1/((x1*0.7+0.3)*(x2*0.7+0.3)))
# }
# NOT RUN {
#Fit the CGP model
mod<-CGP(X,yobs)
summary(mod)
# }