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.
cort is Now on CRAN! You can install the stable version with:
The upstream development version can also be installed with :
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.
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