likelihood.initials: Initial GEV or GP parameters

In theGreatWhiteShark/climex: Estimating Extreme Events in Climatic Time Series (using Data of the DWD)

Initial GEV or GP parameters

Description

Estimates the initial GEV or GP parameters of a time series required to start the fitting routine.

Usage

likelihood.initials(x, model = c("gev", "gpd"), use.skewness = TRUE)

Arguments

x	Time series of class xts or numeric.
model	String determining whether to calculate the initial parameters of the GEV ("gev") or GP ("gpd") function. Default = "gev"
use.skewness	Determines if the skewness is getting used to determine the initial shape parameter. Default = TRUE.

Details

Two main methods are used for the estimation: the L-moments method of Hosking & Wallis and an estimation using the first two moments of the Gumbel distribution. For the later one a modification was added: By looking at skewness of the time series x and with respect to some heuristic thresholds a shape parameter between -.4 and .2 is assigned for the GEV distribution. In case of the GP one, the sign of the skewness matches the sign of the series' shape parameter.

Warning: both methods do not work for samples with diverging (or pretty big) mean or variance.

If no working initial parameter combination could be found using those methods, the function will perform a constrained random walk on the parameters until a working pair is found.

Value

Numerical vector containing the c( location, scale, shape ) estimates for method = "gev" or the c( scale, shape ) estimates for method = "gpd".

Author(s)

Philipp Mueller

See Also

Other optimization: fit.gev.default, fit.gev.list, fit.gev.xts, fit.gev, fit.gpd.default, fit.gpd.list, fit.gpd.xts, fit.gpd, likelihood.augmented, likelihood.gradient.augmented, likelihood.gradient, likelihood

theGreatWhiteShark/climex documentation built on July 13, 2022, 9:11 a.m.