Extremal Index Estimation
Estimation of the extremal index using interexceedance times.
fitexi(data, threshold)
data |
A matrix with two columns: |
threshold |
The threshold. |
The extremal index estimator proposed by Ferro and Segers (2003) is based on interexceedance times. In particular, it does not require a specific declusterization of the time series.
The tim.cond gives an “automatic” procedure to decluster
the time series without any subjective choice to define the
independence condition between clusters.
This function returns a list with two components. The first one
exi gives the estimate of the extremal index; while the
second, tim.cond gives the time condition for independence
between events to be passed to the clust function.
Mathieu Ribatet
Ferro, C. and Segers, J. (2003) Inference for clusters of extreme values. Journal of the Royal Statistical Society. Series B 65:2 545–556.
n.obs <- 500 x <- rexp(n.obs + 1) y <- pmax(x[-1], x[-(n.obs + 1)])## The extremal index is 0.5 u <- quantile(y, 0.95) fitexi(y, u)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.