plgp (version 1.1-12)

pred.GP: Prediction for GPs

Description

Prediction on a per-particle basis for Gaussian process (GP) regression, classification, or combined unknown constraint models

Usage

pred.GP(XX, Zt, prior, Y = NULL, quants = FALSE, Sigma = FALSE,
        sub = 1:Zt$t)
pred.CGP(XX, Zt, prior, mcreps = 100, cs = NULL)
pred.ConstGP(XX, Zt, prior, quants = TRUE)

Value

A single-row data.frame is returned with the desired predictive; these rows are automatically combined when used with

papply

Arguments

XX

matrix or data.frame containing (a design of) predictive locations where ncol(XX) = ncol(X), on which the data were trained and particle Zt thus obtained

Zt

the particle describing model parameters and sufficient statistics that determines the predictive distribution

prior

prior parameters passed from PL generated by one of the prior functions, e.g., prior.GP

Y

not for external use; used internally by CGP and ConstGP internal routines

quants

a scalar logical indicating if predictive quantiles should be are desired

Sigma

a scalar logical indicating if the full predictive variance-covariance matrix is desired; typically only used internally by CGP and ConstGP

sub

not for external used; used internally by CGP and ConstGP internal routines

mcreps

number of Monte Carlo iterations used in CGP prediction, integrating over the latent real-valued Y variables at the XX locations

cs

indicates a class label at which the predictive probability is desired; the entire probability distribution over all class labels will be provided if not specified

Author

Robert B. Gramacy,

Details

For pred.GP the predictive mean (and quantiles if quants = TRUE is provided. For pred.CGP the predictive distribution over the class labels is provided, unless only one class (cs) is desired. pred.ConstGP is a combination of the pred.GP and pred.CGP methods

It is suggested that this function is used in as an argument to papply to obtain many predictions - one for each particle in a cloud - which are combined into a data.frame

Some of the function arguments aren't meant to be specified by the user, but are rather there to facilitate usage as a subroutine inside other PL functions, such as lpredprob.GP and others

References

Gramacy, R. and Polson, N. (2011). “Particle learning of Gaussian process models for sequential design and optimization.” Journal of Computational and Graphical Statistics, 20(1), pp. 102-118; arXiv:0909.5262

Gramacy, R. and Lee, H. (2010). “Optimization under unknown constraints”. Bayesian Statistics 9, J. M. Bernardo, M. J. Bayarri, J. O. Berger, A. P. Dawid, D. Heckerman, A. F. M. Smith and M. West (Eds.); Oxford University Press

Gramacy, R. (2020). “Surrogates: Gaussian Process Modeling, Design and Optimization for the Applied Sciences”. Chapman Hall/CRC; https://bobby.gramacy.com/surrogates/

https://bobby.gramacy.com/r_packages/plgp/

See Also

papply, PL, lpredprob.GP

Examples

## See the demos via demo(package="plgp") and the examples
## section of ?plgp