plot.FEXP {longmemo}

Plot Method for FEXP and WhittleEst Model Fits

Description

(S3) methods for the generic functions plot (and lines) applied to fractional EXP (FEXP) and "WhittleEst" (FEXPest, models. plot() plots the data periodogram and the ‘FEXP'’ model estimated spectrum, where lines() and does the latter.

Usage

## S3 method for class 'FEXP'
plot(x, log = "xy", type = "l",
      col.spec = 4, lwd.spec = 2, xlab = NULL, ylab = expression(hat(f)(nu)),
      main = paste(deparse(x$call)[1]), sub = NULL, ...)

## (With identical argument list:)
## S3 method for class 'WhittleEst'
plot(x, log = "xy", type = "l",
      col.spec = 4, lwd.spec = 2, xlab = NULL, ylab = expression(hat(f)(nu)),
      main = paste(deparse(x$call)[1]), sub = NULL, ...)

## S3 method for class 'FEXP'
      lines(x, type = "l", col = 4, lwd = 2, ...)
## S3 method for class 'WhittleEst'
lines(x, type = "l", col = 4, lwd = 2, ...)

Arguments

x

an R object of class "FEXP", as from FEXPest().

log

character specifying log scale should be used, see plot.default. Note that the default log-log scale is particularly sensible for long-range dependence.

type

plot type for the periodogram, see plot.default.

col.spec, lwd.spec, col, lwd

graphical parameters used for drawing the estimated spectrum, see lines.

xlab, ylab, main, sub

labels for annotating the plot, see title, each with a sensible default.

...

further arguments passed to plot.default.

Author(s)

Martin Maechler

See Also

FEXPest, WhittleEst; plot.default and spectrum.

Examples

data(videoVBR)
fE <- FEXPest(videoVBR, order = 3, pvalmax = .5)
plot(fE)
fE3 <- FEXPest(videoVBR, order = 3, pvalmax = 1)#-> order 3
lines(fE3, col = "red3", lty=2)

f.GN    <- WhittleEst(videoVBR)
f.am21  <- WhittleEst(videoVBR, model = "fARIMA",
                      start= list(H= .5, AR = c(.5,0), MA= .5))
lines(f.GN,   col = "blue4")
lines(f.am21, col = "goldenrod")

##--- Using a tapered periodogram ---------
spvVBR <- spec.pgram(videoVBR, fast=FALSE, plot=FALSE)
fam21 <- WhittleEst(periodogr.x = head(spvVBR$spec, -1),
                    n = length(videoVBR), model = "fARIMA",
                    start= list(H= .5, AR = c(.5,0), MA= .5))
fam21
f.am21 # similar but slightly different

plot(fam21)

## Now, comparing to traditional ("log-X", not "log-log") spectral plot:
plot(fam21, log="y")

## compared to the standard R spectral plot : s
if(dev.interactive(TRUE)) getOption("device")()# a new graphics window
plot(spvVBR, log = "yes", col="gray50")
all.equal(.ffreq(fE$n) / (2*pi) -> ffr.,
          head(spvVBR$freq, -1))# TRUE
lines(ffr., fam21$spec, col=4, lwd=2)
## need to adjust for different 'theta1':
lines(ffr., f.am21$spec * fam21$theta1 / f.am21$theta1,
      col = adjustcolor("tomato", 0.6), lwd=2)