Fits univariate extreme value distributions to data
These functions fit the generalised extreme value and generalised Pareto distribution to data using maximum likelihood.
gevmle(x, ..., method = "Nelder") gpdmle(x, threshold, ..., method = "Nelder")
x |
Numeric vector of observations |
... |
Optional arguments to be passed to the
|
threshold |
Numeric. The threshold value. |
method |
The numerical optimisation method to be used. |
These two functions are “extremely light” functions to fit the
GEV/GPD. These functions are mainly useful to compute starting values
for the Schlather and Smith model - see fitmaxstab.
If more refined (univariate) analysis have to be performed, users should use more specialised packages - e.g. POT, evd, ismev, ....
A vector for the estimated parameters of the GEV/GPD.
Mathieu Ribatet
## 1 - GEV fit x <- rep(NA, 100) for (i in 1:100) x[i] <- max(rnorm(365)) gevmle(x) ## 2- GPD fit x <- rnorm(10000) ##we need to fix a threshold u <- quantile(x, 0.99) gpdmle(x, u)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.