plot.pgrid {ExceedanceTools}

Plots pgrid object.

Description

plot.pgrid plots a grid of pixels based on a pgrid object.

Usage

## S3 method for class 'pgrid'
plot(x, set, col = "gray", add = FALSE, type = "confidence", ...)

Arguments

x

An pgrid object returned from the pgrid function.

set

A vector which contains the indices of the pixels/cells that should be plotted. OR a confreg object from the confreg function. See Details.

col

The color of the plotted pixels.

add

A logical value indicating whether the pixels should be added to an existing plot (add = TRUE) or should the pixels be plotted on a new plot (add = FALSE).

type

The type of set of plot if set of of class confreg. The default is "confidence", while the other option is complement, based on the components of the confreg object.

...

Additional arguments that will be passed to the image function (assuming add=FALSE).

Details

If a vector of pixel indices is supplied to set, then those pixels will be colored col by this function and the type argument has no effect. On the other hand, if the set argument is of class confreg, then the function digs in to display either the confidence or complement set in the confreg object. In that case, type is used to decide which set to display.

Value

This function does not return anything; it only creates a new plot or modifies an existing plot.

Author(s)

Joshua French

Examples

library(SpatialTools)

# Example for exceedance regions

set.seed(10)
# Load data
data(sdata)
# Create prediction grid
pgrid <- create.pgrid(0, 1, 0, 1, nx = 26, ny = 26)
pcoords <- pgrid$pgrid
# Create design matrices
coords = cbind(sdata$x1, sdata$x2)
X <- cbind(1, coords)
Xp <- cbind(1, pcoords)

# Generate covariance matrices V, Vp, Vop using appropriate parameters for 
# observed data and responses to be predicted
spcov <- cov.sp(coords = coords, sp.type = "exponential", 
 sp.par = c(1, 1.5), error.var = 1/3, finescale.var = 0, pcoords = pcoords)

# Predict responses at pgrid locations
krige.obj <- krige.uk(y = as.vector(sdata$y), V = spcov$V, Vp = spcov$Vp, 
 Vop = spcov$Vop, X = X, Xp = Xp, nsim = 100, 
 Ve.diag = rep(1/3, length(sdata$y)) , method = "chol")
                
# Simulate distribution of test statistic for different alternatives
statistic.sim.obj.less <- statistic.sim(krige.obj = krige.obj, level = 5, 
 alternative = "less")
statistic.sim.obj.greater <- statistic.sim(krige.obj = krige.obj, 
 level = 5, alternative = "greater")
# Construct null and rejection sets for two scenarios
n90 <- exceedance.ci(statistic.sim.obj.less, conf.level = .90, 
 type = "null")
r90 <- exceedance.ci(statistic.sim.obj.greater,conf.level = .90, 
 type = "rejection")       
# Plot results
plot(pgrid, n90, col="blue", add = FALSE, xlab = "x", ylab = "y")
plot(pgrid, r90, col="orange", add = TRUE)
legend("bottomleft", 
 legend = c("contains true exceedance region with 90 percent confidence", 
 "is contained in true exceedance region with 90 percent confidence"),
 col = c("blue", "orange"), lwd = 10)