Extremal Dependence Models

A set of procedures for parametric and non-parametric modelling of the dependence structure of multivariate extreme-values is provided. The statistical inference is performed with non-parametric estimators, likelihood-based estimators and Bayesian techniques. It adapts the methodologies of Beranger and Padoan (2015) , Marcon et al. (2016) , Marcon et al. (2017) , Marcon et al. (2017) and Beranger et al. (2021) . This package also allows for the modelling of spatial extremes using flexible max-stable processes. It provides simulation algorithms and fitting procedures relying on the Stephenson-Tawn likelihood as per Beranger at al. (2021) .

ExtremalDep: Extremal Dependence Models

A set of procedures for parametric and non-parametric modelling of the dependence structure of multivariate extreme-values is provided. The statistical inference is performed with non-parametric estimators, likelihood-based estimators and Bayesian techniques. It adapts the methodologies of Beranger and Padoan (2015) doi:10.48550/arXiv.1508.05561, Marcon et al. (2016) doi:10.1214/16-EJS1162, Marcon et al. (2017) doi:10.1002/sta4.145, Marcon et al. (2017) doi:10.1016/j.jspi.2016.10.004 and Beranger et al. (2021) doi:10.1007/s10687-019-00364-0. This package also allows for the modelling of spatial extremes using flexible max-stable processes. It provides simulation algorithms and fitting procedures relying on the Stephenson-Tawn likelihood as per Beranger at al. (2021) doi:10.1007/s10687-020-00376-1.

Version 0.0.4-2 updates:

• Fixes a typo in dExtDep() for in the Asymetric Logistic model;
• Replaces the Calloc() and Free() calls in the .C files by the R_* prefixed counterparts since STRICT_R_HEADERS=1 becomes the default with R 4.5.0;
• Improves some entries in the manual.

Version 0.0.4-3 updates:

• Inclusion of the PAMfmado() function written by Philippe Naveau.

Version 0.0.4-4 updates:

• Fixes resetting graphical parameters in plot_ExtDep.np() when type = "Qsets";
• Replaces closeAllConnections() by stopCluster() in fExtDepSpat().

Version 0.0.4-5 updates:

• Improvement to dExtDep() function when model="HR" and "ET";
• New lambda.hr() function that can be used to define the parameters of the trivariate Husler-Reiss model. Given two parameters, a range for the third parameter is provided to ensure positive definite matrices in the exponent function;
• Improve efficiency when manipulating matrices:
  • t(A) %*% B is replaced by crossprod(A,B) (and vice versa);
  • solve(A) is replaced by chol2inv(chol(A));
  • t(x) %*% solve(A) %*% x is replaced by sum(forwardsolve(t(chol(A)),x));
• Define outputs of fExtDep(), fExtDep.np() and fExtDepSpat() to be of class ExtDep_Freq, ExtDep_Bayes, ExtDep_npFreq, ExtDep_npBayes, ExtDep_npEmp and ExtDep_Spat;
• Make use of class methods for plot() and summary() functions replacing the previous plot.ExtDep(), etc;
• New est(), StdErr(), logLik(), bic(), tic() to extract outputs from objects of class ExtDep_Freq, ExtDep_Bayes and ExtDep_Spat.

Version 1.0.0 updates:

• Version number has been updated to reflect the regular construction ...
• Improved formatting across all .R files using the styler package.
• Improved formatting of all manual (.Rd) files to follow the tidyverse style guide (spacing, indentation, line width, ...).