log_likelihood_generalized_2d: Generalized Log-Likelihood Function for 2D Copula-GEV Model

In STCCGEV: Conditional Copula Model for Crop Yield Forecasting

Generalized Log-Likelihood Function for 2D Copula-GEV Model

Description

Computes the negative log-likelihood of a 2-dimensional copula-GEV model, incorporating dynamic Generalized Extreme Value (GEV) parameters and a time-varying copula structure.

Usage

log_likelihood_generalized_2d(params, u1, u2, X_t, z1, z2, copula)

Arguments

params	Numeric vector, model parameters including copula and GEV parameters.
u1	Numeric vector (length n_train), pseudo-observations for margin 1.
u2	Numeric vector (length n_train), pseudo-observations for margin 2.
X_t	Numeric matrix (⁠n_train x M⁠), risk factors affecting copula parameters.
z1	Numeric matrix (⁠n_train x M⁠), observed data for margin 1.
z2	Numeric matrix (⁠n_train x M⁠), observed data for margin 2.
copula	Character, specifying the copula type: "Clayton", "Frank", "Gumbel", "Joe", or "Gaussian".

Value

The negative log-likelihood value for optimization.

Examples

test_ll_2d <-log_likelihood_generalized_2d(init_params_full,
                                  uu[,1],
                                  uu[,2],
                                  xx_train,
                                  zz_train[,1,],
                                  zz_train[,2,],
                                  "Gaussian")

STCCGEV documentation built on April 4, 2025, 1:50 a.m.