Northop and Coleman piecewise generalized Pareto threshold selection diagnostic

The model tests the null hypothesis of a generalized Pareto above each threshold in thresh against the alternative of a piecewise generalized Pareto model with continuity constraints.

Usage

thselect.ncpgp(xdat, thresh, test = "score", plot = FALSE, level = 0.95, ...)

Arguments

xdat

[vector] observations

thresh

[vector] candidate thresholds

test

[string] indicating whether to perform score test or likelihood ratio (lr) test. The latter requires fitting the alternative model, and so is more computationally expensive.

plot

[logical]; if TRUE, return a plot with the p-value path.

level

[double] confidence level for confidence interval, defaults to 0.95

...

additional arguments, for backward compatibility purposes

Value

an object of class mev_thselect_ncpgp containing the test statistic (stat), the p-values (pval), the threshold candidates (thresh) and the selected threshold (thresh0).