extRemes (version 1.65)

gen.gev: Simulate Data from a Generalized Extreme Value (GEV) Distribution

Description

Generates data from a GEV (GPD) distribution function. May also incorporate a linear trend in location parameter of GEV.

Usage

gen.gev(p, n, trend = NULL)
gen.gpd(n,sigma,xi,u)

Arguments

p

A length 3 vector indicating the mean, scale and shape of the GEV, respectively.

n

The sample size to generate.

trend

Slope of the location parameter trend (if desired).

sigma

Scale parameter of GPD.

xi

Shape parameter of GPD.

u

Threshold for GPD.

Value

• Returns a vector of simulated data.

Details

Value returned (with no trend) is derived from the follwing formula (GEV).

mu + sigma (X**(-xi - 1))*xi,

where X is a uniform random variable.

For GPD the formula is:

sigma/xi * ((1-runif(n))**(-xi-1) for xi != 0 and

rexp(n, rate=1/sigma) for xi = 0.

References

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

See Also

From ismev package: gev.fit, gev.diag, gpd.fit, gpd.diag

Examples

# obtain a GEV with mean, 4, scale 1.5 and shape of -0.1
mu <- 4 # location parameter
sigma <- 1.5 # scale parameter
xi <- -0.1 # shape parameter

params <- c( mu, sigma, xi)

# generate a sample of size 25
gen1 <- gen.gev( p=params, n=25)

# Now generate one with a trend.
gen2 <- gen.gev( p=params, n=25, trend=0.1)

# Fit 'gen1' to a GEV distribution and plot the diagnostics.
gen1.fit <- gev.fit( gen1)
class( gen1.fit) <- "gev.fit"
plot( gen1.fit)

# Fit 'gen2' to a GEV distribution and plot the diagnostics.
gen2.fit1 <- gev.fit( gen2)
class( gen2.fit1) <- "gev.fit"
plot( gen2.fit1)