gpd.diag: Diagnostic plots for a fit to the Generalized Pareto...

In K-Molloy/tevt: Threshold Methods for Extreme Value Theory

Description

Allows calling of a 2x2 diagnostic plot, different plots available for stationary vs non-stationary models. The extension from EVA allows for direct calling to gpd.diag and plots produced. Additionally, it exposes the individual plots for custom sizing (especially via .rmd or .rnw)

Usage

gpd.diag(z, conf = 0.95, method = c("delta", "profile"), safe = FALSE)

Arguments

z	A class object returned from ‘gpdFit’.
conf	Confidence level used in the return level plot.
method	The method to compute the return level confidence interval - either delta method (default) or profile likelihood. Choosing profile likelihood may be quite slow.
safe	Binary (default FALSE) For when plot margins are too large

Details

See the reference for details on how return levels are calculated.

Value

For stationary models, provides return level, density, probability, and quantile plots for the GPD exceedances. The overlaid density is the ‘true’ density for the estimated parameters. For nonstationary models, provides residual probability and quantile plots. In addition, nonstationary models provide plots of the residuals vs. the parameter covariates.

References

Coles, S. (2001). An introduction to statistical modeling of extreme values (Vol. 208). London: Springer.

Modified from "eva : Extreme Value Analysis with Goodness-of-Fit Testing."

Examples

## Not run: 
x <- rgpd(10000, loc = 0.5, scale = 1, shape = 0.1)
z <- gpd.fit(x, nextremes = 500)
plot(z)

## End(Not run)

K-Molloy/tevt documentation built on Dec. 18, 2021, 2:34 a.m.