Some Empiric and Nonparametric Copula Models

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 proposed by Laverny, Maume-Deschamps, Masiello and Rullière (2020) , including the localised dimension reduction, which fits a copula by recursive splitting of the copula domain. We also 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 copula.