plot.mcpot {POT}

Graphical Diagnostics: Markov Chains for All Exceedances.

Description

Plot several graphics to judge goodness of fit of the fitted model.

Usage

## S3 method for class 'mcpot'
plot(x, opy, exi, mains, which = 1:4, ask = nb.fig <
length(which) && dev.interactive(), acf.type = "partial", ...)

Arguments

x

An object of class "bvpot". Most often, the object returned by the fitbvgpd function.

opy

Numeric. The number of Observation Per Year (or more generally per block). If missing, the function warns and set it to 365.

exi

Numeric. The extremal index value. If missing, the estimator of Ferro and Segers (2003) is used.

mains

May be missing. If present a 4–vector of character strings which gives the titles of the plots.

which

a numeric vector which specifies which plot must be drawn: '1' for the auto correlation plot, '2' for Pickands' Dependence Function plot, '3' for the spectral density plot and '4' for a bivariate return level plot.

ask

Logical. If TRUE, user is asked before each plot.

acf.type

The type of auto correlation to be plotted. Must be one of "correlation", "covariance" or "partial" (the default). See the acf function.

...

Other parameters to pass to the plot function.

Value

Several plots and returns invisibly the return level function.

Warning

See the warning for the return level estimation in documentation of the retlev.mcpot function.

Note

For the return level plot, the observations are not plotted as these are dependent realisations. In particular, the return periods computed using the prob2rp are inaccurate.

Author(s)

Mathieu Ribatet

References

Ferro, C. and Segers, J. (2003). Inference for clusters of extreme values. Journal of the Royal Statistical Society B. 65: 545–556.

See Also

fitmcgpd, acf, retlev

Examples

set.seed(123)
mc <- simmc(200, alpha = 0.5)
mc <- qgpd(mc, 0, 1, 0.25)
Mclog <- fitmcgpd(mc, 1)
par(mfrow=c(2,2))
rlMclog <- plot(Mclog)
rlMclog(T = 3)