Estimate the extremal index (and confidence limits) using intervals estimation.
Usage
eiAnalyze(x, thresholds = quantile(x, seq(0.9, 0.995, by = 0.005)), conf = 0.95, iter = 500, plot=FALSE)
Arguments
x
'n X 1' vector of data observations.
thresholds
Single number, or 'm X 1' vector of thresholds above which to estimate the extremal index.
conf
The conf*100 percent confidece level.
iter
Number of iterations desired for bootstrapping.
plot
Logical whether or not to plot the extremal indices against the thresholds (w/ confidence bounds).
Only used if the length of 'thresholds' is greater than 1.
Value
Returns a list with the following components.
ei'm X 1' vector of the estimate(s) of the extremal index.
ciAn 'm X 2' matrix giving the conf*100 percent confidence limits for ei.
uThe threshold(s) used.
ncAn 'n X 1' vector giving the number of clusters for each threshold choice.
run.lengthAn 'm X 1' vector giving the estimated run lengths.
Details
Uses several functions written by Chris Ferro for estimating the extremal index by intervals estimation
(Ferro and Segers (2003)). Specifically, it calls the functions: exi.intervals, decluster.intervals,
boot.matrix and boot.sequence.
References
Ferro CAT and Segers J (2003) Inference for clusters of extreme
values. Journal of the Royal Statistical Society B 65, 545-556.