gpdThresh: Produce plots to aid threshold selection for GPD models
In harrysouthworth/margarita: Support functions for texmex

Description Usage Arguments Details Value

Produce plots to aid threshold selection for GPD models

gpdThresh(
  x,
  umin = quantile(x, 0.05),
  umax = quantile(x, 0.95),
  nint = 25,
  which = 1:2,
  priorParameters = NULL,
  cov = "observed"
)

`x`	A numeric vector.
`umin`	The minimum value of x to use as a threshold. Defaults to `umin=quantile(x, .05)`.
`umax`	The maximum value of x to use as a threshold. Defaults to `umin=quantile(x, .95)`.
`nint`	The number of values of x at which to compute the parameters. Defaults to `nint=25`.
`which`	Which plots to produce. If `which` contains 1, the mean residual life plot is included; if it contains 2, the plots of the generalized Pareto parameters over increasing thresholds are included; if it includes 3, the plot of the EGP3 power parameter is also included. Defaults to including all 3.
`priorParameters`	A list containing the parameters of the prior distribution, passed through to `evm`.
`cov`	How to compute the covariance, passed through to `evm`. Defaults to `cov='observed'`, but you might want to use `cov='numeric'` if the computations are unstable.
`...`	Additional arguments passed to `ggplot`.

The use of egp3 was experimental. Experience suggests it's better to think carefully about where a sensible threshold might be before using egp3. Note that the dimension of the parameter vector for egp3 is greater than that for gp so that if you use non-NULL prior parameters, the function can't cope. If you really want to use prior parameters, call the functions separately (egp3RangeFit, mrl, gpdRangeFit).

A list of graphical objects created by ggplot.

harrysouthworth/margarita documentation built on Aug. 19, 2021, 5 a.m.