L1O: Leave-one-out diagnostics for Gaussian process

In mcrucifix/gp: Gaussian process calibration for emulation

Leave-one-out diagnostics for Gaussian process

Description

Provides leave-one-out diagnostics for Gaussian process given a design X with data Y and hyperparameters lambda.

Usage

L1O(X, Y, lambda, ...)

Arguments

X	Matrix of n lines corresponding to experiment members, and with m columns corresponding to different inputs. Can be one column, but always need to have this matrix form (use as.matrix if needed )
Y	Currently the Gaussian process is univariate. So, vector with n elements corresponding to outputs.
lambda	List made of (1) a vector codetheta, with $m$ elements corresponding to roughness lengths associated with input variables, and (2)
...	Other parameters passed to GP_C

Value

Vector of

Xi	Input design
Yi	Supplied output
mean	GP output mean for each left-out experiment
sd	GP standard error for each left-out experiment
mahalanobis	GP mahalanobis for each left-out experiment

Author(s)

Michel Crucifix

See Also

GP_C, GP_P

Examples

 # univariate example
 X <- matrix(c(1,2,3,4,5,6,7), 7, 1)
 Y <- c(1.1, 2.1, 4.7, 1.3, 7.2, 8, 9)
 
 out <- L1O(X, Y, lambda=list(theta=1, nugget=0))
 
 require(Hmisc)

 with(out, errbar(Yi, mean, mean-sd, mean+sd))
 lines(c(-10,10),  c(-10,10), lty=2)

mcrucifix/gp documentation built on July 29, 2023, 8:58 p.m.