The cort package provides S4 classes and methods to fit several copula models:

• The classic empirical checkerboard copula and the empirical checkerboard copula with known margins, see Cuberos, Masiello and Maume-Deschamps (2019) are proposed. These two models allow to fit copulas in high dimension with a small number of observations, and they are always proper copulas. Some flexibility is added via a possibility to differentiate the checkerboard parameter by dimension.

• The last model consist of the implementation of the Copula Recursive Tree algorithm, aka. CORT, including the localised dimension reduction, which fits a copula by recursive splitting of the copula domain, see Laverny, Maume-Deschamps, Masiello and Rullière (2020).

• We finally provide an efficient way of mixing copulas, allowing to bag the algorithm into a forest, and a generic way of measuring d-dimensional boxes with a given copula.

Installation

cort is Now on CRAN! You can install the stable version with:

install.packages("cort")

The upstream development version can also be installed with :

devtools::install_github("lrnv/cort")

The vignettes are quite expressive. They give a clear overview of what can be done with this package, how it is coded and why it is useful. Please read them for more details.

References

Cuberos A, Masiello E, Maume-Deschamps V (2019). “Copulas Checker-Type Approximations: Application to Quantiles Estimation of Sums of Dependent Random Variables.” Communications in Statistics - Theory and Methods, 1--19. ISSN 0361-0926, 1532-415X.

Laverny O, Maume-Deschamps V, Masiello E, Rullière D (2020). “Dependence Structure Estimation Using Copula Recursive Trees.” arXiv preprint arXiv:2005.02912