Create families of distributions for use with extreme value modelling.
texmexFamily(name, log.lik, param, info = NULL, sandwich =
NULL, start = NULL, resid = NULL, rl, delta, endpoint, density,
rng, prob, quant)
# S3 method for texmexFamily
print(x,...)
# S3 method for texmexFamily
summary(object,...)
# S3 method for summary.texmexFamily
print(x,...)
A object of class "texmexFamily", which is essentially a
list containing the input arguments. If info
,
sandwich
, start
, resid
are not provided,
they default to NULL
.
The name of the distribution.
The distribution's log-likelihood function.
The names of the parameters in the model.
Function to compute the information matrix. If not provided, the modelling functions will work with a numerical approximation.
Function to compute the filling in the Huber
sandwich estimator of the covariance matrix of parameter
estimates, used for dependent data. Only implemented in family
gpd
.
Function to compute starting parameters for the model. If not provided, the modelling functions will try to guess.
Function to compute residuals for the model.
Function to compute return levels.
Function to compute adjustments for covariance for return levels.
Function to compute the upper or lower endpoint of the fitted distribution.
Function to compute the density.
Function for random number generation.
Function to compute cumulative probabilities.
Function to compute quantiles.
Additional arguments to the print and summary methods.
An object of class 'texmexFamily'.
Harry Southworth
The density
, rng
, prob
and quant
functions can be simple wrappers for the usual d, r, p and q
functions. They should take a matrix with number of columns equal
to the number of parameters, and a fitted model object even if the
model object is not used by the function.
Examples of "texmexFamily" objects are gpd
, gev
, glo
,
gpdIntCensored
, weibull
, gumbel
and egp3
. Take a look at
those objects to see how the functions should be constructed.
The functions are used by the modelling functions to create diagnostic plots, predictions, etc..
evm