BLOCK MAXIMA ESTIMATION

Share:

Description

Divide the vector data into (possibly overlapping) blocks and estimate the shape parameter of the upper tail with by the GEV shape parameter of the sample of block maxima.

Usage

1
block.max(data, overlap = 0, nb.blocks = NA, block.size = 100, method = "ml")

Arguments

data

Numerical vector representing the sample observations

overlap

Integer giving the size of the overlap between successive blocks

nb.blocks

Number of blocks used

block.size

Integer giving the length of any given block

method

String which should be either "ml" or "lmom" depending upon the estimation method used.

Details

The overlap given by the parameter overlap should be no greater than 50. The block length given by the parameter block.size should be at least 100. The estimation method given by the parameter method should be either ml for maximum likelihood (default) or lmom for the method of moments.

Value

Returns a list comprising

n

Length of the original data vector

data

ordered original data

method

the method used for the estimation

nb.blocks

Number of blocks used

overlap

Size of the overlap

intervl.a

vector of the left hand points of the intervals used in the estimation

intervl.b

vector of the right hand points of the intervals used in the estimation

param.est

esimate of the tail shape parameter

Author(s)

Rene Carmona

References

Rene Carmona: Heavy Tail Distributions and Financial Applications. Springer Verlag (2008)