mev version 2.1

mev 2.1

New:

• Function thselect.egp for threshold selection with tests similar to Northop and Coleman routine.
• Function thselect.cbm for threshold selection with semiparametric bootstrap mean square error estimation of shape parameter.

Changes:

• thselect.vmetric argument ci has been renamed to level (backward compatible).
• Arguments method for tstab.egp and tstab.gp now has "lrt" instead of "profile" (choice backward compatible).
• thselect.sdinfo gains an argument k to select the number of gaps, and returns a threshold thresh0 and k as part of the object output. The print method has been modified to reflect this.

Fixes:

• Function pegp did not work, and did not correctly handle lower.tail argument.
• Functions degp and regp correctly handle the case shape=0 for models pt-beta and pt-gamma.
• thselect.mdps now return shape parameter (rather than reciprocal shape), consistent with other semiparametric methods
• thselect.pickands now correctly handles user-provided threshold vectors.
• Warnings are not produced when fit.egp fails to obtain standard errors.
• The print method for thselect.egp is more explicit about the lack of threshold returned if the hypothesis is rejected at the highest threshold considered (which returns NA).
• thselect.ncpgp now passes ordered thresholds to inner function to avoid warning messages.
• Function shape.hill now provides more informative error messages when passing vectors of invalid number of exceedances k.

mev version 2.0

Major release with breaking changes (mostly function names and arguments) for more coherence. Many functions now flagged as deprecated, but changes to the syntax are often backward compatible.

New:

• All functions producing plot now export objects with associated S3 routines.
• Stein's weighted generalized Pareto estimator with fit.wgpd and suggested weights Stein_weights.
• Multiple shape estimators (Hill, Pickands, moment estimator, generalized quantile) via fit.shape and specific routines shape.hill, shape.erm, shape.genjack, shape.lthill, shape.trunhill, shape.rbm, shape.moment, shape.pickands, shape.osz, shape.genquant and shape.vries.
• Second order regular variation parameter estimation via fit.rho (four estimators available, via rho.dk, rho.fagh, rho.gbw and rho.ghp).
• L-moment estimator for generalized Pareto distribution via gpd.lmom.
• Estimator of tail dependence of Krupskii and Joe (kjtail) and through xdep.eta.
• Weissman quantile estimator (qweissman)
• Function dgeoaniso for geometric anisotropy, using the parametrization of Rai and Brown (2025)
• Function maxstable for changes of parameters for GEV.
• Distributions for extended generalized Pareto (d/p/q/r egp), with new models and profile likelihood (egp.pll). This also affects the plotting routines. The nomenclature has been modified for argument 'model' in egp.fit, egp.ll and egp.retlev with backward compatibility.
• New methods for threshold selection (16 in total) thselect.alrs, thselect.bab, thselect.cv, theselect.expgqt, thselect.gbw, thselect.mdps, thselect.mrl (which supersedes automrl), thselect.pec, thselect.pickands, thselect.rbm, thselect.samsee,
• Threshold stability plots for different estimators tstab.hill, tstab.lthill (with different diagnostics), tstab.mrl

Changes:

• Function PickandsXU is deprecated and replaced with shape.osz
• Function fit.gpd with method obre now returns ordered exceedances and weights.
• spunif now correctly handles missing values.
• d/p/q/r for gev and gp now fully vectorized with respect to arguments, including shape parameter
• confint.eprof has an argument for boundary non-regular null distribution (for the most common case).
• Breaking changes to fit.egp, egp.retlev, tstab.egp. The former now allows fixed parameters (for profiling).
• All functions for threshold selection are now named using the prefix thselect. Their arguments are somewhat standardized, and they each have distinct plot and print methods with automatic selection.
• All threshold stability plots now have the same syntax, with prefix tstab.
• Measures of extremal dependence now via functions prefixed with xdep., including xdep.chi, xdep.eta, xdep.chibar and xdep.asym. The first three call taildep, the last one supersedes xasym.
• Function lambdadep deprecated, replaced with adf to better comply with current nomenclature.
• Function automrl deprecated in favour of thselect.mrl.
• Function chibar removed in favour of xdep.chibar.
• Arguments of extcoef for marginal transformation margtrans have been modified (backward compatible changes).
• Function fit.egp now includes alternative arguments to select the optimization routine
• Internal function (exported, but hidden) jac has been replaced by jac_gpd_pareto to avoid ambiguity
• Tests for max stability (maxstabtest) and independence via score (scoreindep) now renamed test.maxstab and test.scoreindep
• Change to name infomat.test now thselect.sdinfo, vmetric now thselect.vmetric, W.diag now thselect.wseq, NC.diag now thselect.ncpgp
• Arguments of xasym modified to make syntax more uniform.
• Function mvrnorm renamed to rmvnorm.

Fixes:

• Function adf now allow for values above 1 for bivariate angular dependence function.
• Function .infomat needs not have data argument.
• Fix generalized Pareto fit with fixed parameters

Version 1.17

mev 1.17 (Release date 2024-07-09)

Fixes:

• gpd.ll and gpd.nll correctly return a non-NA value when xi=-1 (#18, reported by Paddy O'Toole)
• tstab.gpd now correctly works when input leads to boundary values, and profile likelihood intervals are now correctly truncated in the range of admissible values
• fit.gpd now relies on gpd.ll to return the negative log likelihood value nllh
• Threshold stability plots now compute limits of plot safely, handling missing values in parameter estimates. Intervals are truncated to admissible values
• For multivariate censored likelihood, marginal thresholds equal to the functional threshold (default option) doesn't return an error.

mev 1.16 (Release date 2023-11-30)

New:

• Pickand's U-statistic estimator of the shape
• New datasets: Newlyn wave height, pandemics, nutrients

Fixes:

• gp.tstab optimization bounds for profile confidence interval fixed.
• rparp failed when a single draw was accepted (due to matrix being cast to vectors)
• fit.extgp now handles model 0 (generalized Pareto), fixing #17

mev v.1.15

New:

• gev and gp distribution functions (d/p/q/r) introduced to remove dependency on evd package.
• Ramos and Ledford score test of independence

Fixes:

• taildep function now correctly check choice of estimators against list of methods
• Fix argument size in likmgp and clikmgp for the logistic model
• Correctly return a covariance matrix with NAs when the matrix is singular or xi=-1

Changes:

• taildep arguments leads to faster calculations with betacop option.
• Replace nloptr dependency by Rsolnp

mev 1.14

==============

New:

• bivariate coefficient of extremal asymmetry
• four new families of max-stable models (pairwise beta, pairwise exponential, weighted Dirichlet and weighted exponential) for rmev, following Ballani and Schlather (2011)
• maximum likelihood estimation routines (fit.gpd, fit.gev, etc.) now accept fixed parameters
• mean residual life plots with weighted least square fit
• coefficient of variation tests for threshold selection
• Varty et al. metric based threshold selection diagnostic
• anova methods for mev_gpd and mev_gev objects
• rmev and rmevspec now accept a distance matrix in place of coordinates for spatial models.
• new datasets
• website with vignettes

Changes

• Functions W.diag and NC.diag now have S3 plot and print methods
• Changes to arguments (backward compatible) to xdat throughout
• Many dependencies used by single functions are now listed in Suggest.

Fixes:

• ext.coef correctly handles arrays with missing values (reported by M. Jousset)
• Optimization method in fit.gev now uses the PWM solution of Hosking (1985) as starting value
• gev.pll now returns confidence intervals for param = "quant" (reported by D. Dupuis)
• Fixes to NHPP order statistics density (returns -Inf outside of domain, also correctly evaluate for boundary case when xi=-1)
• Optimization routines fit.pp, fit.gev and fit.rlarg now return correct MLE when solution lies on boundary (xi=-1) and are more robust to failure (gradients for nlminb return large values rather than NAs which caused the algorithm to stop).
• Grimshaw's algorithm sometimes returned incorrect value because of too low tolerance for eta near zero. Set back to default settings.
• fit.gpd(..., method = "obre") now returns additional failure messages if the algorithm drifts towards infeasible values.
• rparp now correctly handles xi=0
• Extended GP model now has 'step' for discretization, and a valid distribution function that returns real arguments whenever the input is finite (#9)
• W.diag and egp.fitrange include arguments for changing 'par' (#10)
• smith.penult now computes reciprocal hazard and it's derivative on the log scale (when possible) to avoid numerical overflow.

CRAN v 1.13

Update coordinates in Vignette

CRAN release 1.12

v1.12

CRAN 1.12