gek (version 1.2.0)

plot.gekm: Plot Method for a gekm Object

Description

Visualization of the leave-one-out cross-validation results of a gekm model.

Usage

# S3 method for gekm
plot(x, y, main = "Leave-One-Out", ylim = NULL, panel.first = abline(0, 1),
	add = FALSE, reestim = TRUE, scale = FALSE, df = NULL, add.interval = FALSE, 
	level = 0.95, args.arrows = NULL, ...)

Value

Returns the predicted values of the leave-one-out cross-validation invisibly. If add.interval = TRUE, the lower and upper bounds of the confidence intervals are also returned.

Arguments

x

an object of class "gekm".

y

not used.

main

main title for the plot.

ylim

limits for the y-axis.

panel.first

an expression to be evaluated before the actual plot. Default is abline(0, 1).

add

logical. Should results be added to an already existing plot? Default is FALSE.

reestim

logical. Should the regression coefficients be re-estimated? Default is TRUE, see loo.gekm for details.

scale

logical. Should the estimated process variance be scaled? Default is FALSE, see sigma.gekm for details.

df

degrees of freedom for the \(t\) distribution. Default is NULL, see predict.gekm for details.

add.interval

logical. Should confidence intervals be added? Default is FALSE.

level

confidence level for calculating confidence intervals. Default is 0.95.

args.arrows

a list with further arguments to be passed to arrows. Only used if add.interval = TRUE.

...

further arguments to be passed to plot.default or points.

Author

Carmen van Meegen

Details

For further details on the arguments scale see

References

Bachoc, F. (2013). Cross Validation and Maximum Likelihood Estimations of Hyper-parameters of Gaussian Processes with Model Misspecification. Computational Statistics and Data Analysis, 66:55--69. tools:::Rd_expr_doi("10.1016/j.csda.2013.03.016").

Dubrule, O. (1983). Cross Validation of Kriging in a Unique Neighborhood. Mathematical Geology, 15:687--699. tools:::Rd_expr_doi("10.1007/BF01033232").

Martin, J. D. and Simpson, T. W. (2005). Use of Kriging Models to Approximate Deterministic Computer Models. AIAA Journal, 43(4):853--863. tools:::Rd_expr_doi("10.2514/1.8650").

See Also

gekm for fitting a (gradient-enhanced) Kriging model.

loo for leave-one-out cross-validation.

branin for the Branin-Hoo function.

arrows for drawing arrows.

Examples

## 2-dimensional example: Branin-Hoo function

# Generate a grid for training
n <- 4
x1 <- seq(-5, 10, length = n)
x2 <- seq(0, 15, length = n)
x <- expand.grid(x1 = x1, x2 = x2)
y <- branin(x)
dy <- braninGrad(x)
dat <- data.frame(x, y)
deri <- data.frame(dy)

# Fit (gradient-enhanced) Kriging model
km.2d <- gekm(y ~ .^2, data = dat)
gekm.2d <- gekm(y ~ .^2, data = dat, deriv = deri)

# Plot leave-one-out cross-validation results
plot(km.2d)
plot(km.2d, panel.first = grid())
plot(km.2d, panel.first = {grid(); abline(0, 1, col = 8)})
plot(km.2d, add.interval = TRUE)
plot(km.2d, add.interval = TRUE, pch = 16, col = 4)
plot(km.2d, add.interval = TRUE, pch = 16, col = 4,
	panel.first = {grid(); abline(0, 1)}, 
	args.arrows = list(col = 4, length = 0))

plot(km.2d, pch = 1, col = 4, cex = 1.2, lwd = 2)
plot(gekm.2d, pch = 4, col = 2, cex = 1.2, lwd = 2, add = TRUE)
legend("topleft", legend = c("Kriging", "GEK"), col = c(4, 2), pch = c(1, 4), pt.lwd = 2)

par(mfrow = c(1, 2), oma = c(3.6, 3.5, 1.5, 0.2), mar = c(0, 0, 1.5, 0))
res <- plot(km.2d, col = 7, pch = 16, add.interval = TRUE, main = "Kriging",
	scale = TRUE, panel.first = {grid(); abline(0, 1, col = 8)})
res
plot(gekm.2d, col = 3, pch = 16, add.interval = TRUE, scale = TRUE, main = "GEK",
	ylim = range(res), yaxt = "n", panel.first = {grid(); abline(0, 1, col = 8)})
title(main = "Leave-One-Out", outer = TRUE)
mtext(side = 1, outer = TRUE, line = 2.5, "response")
mtext(side = 2, outer = TRUE, line = 2.5, "prediction")