simul.fun.noGEV: Simulate Multivariate Crop Yield Data Using a Generalized Copula-BSTS Model Without GEV Covariates

Description

This function simulates multivariate crop yield data using a time-varying copula combined with Bayesian Structural Time Series (BSTS) models without GEV covariates for comparision.

Usage

simul.fun.noGEV(
  nsim = 100,
  n_train,
  n_test,
  copula,
  init_params,
  fn,
  U_train,
  Z_train,
  Z_test,
  X_train,
  X_test,
  Y_test,
  BSTS_list
)

Value

A list containing:

optim_results: Results from the optimization process.
theta_sim: Simulated copula parameters across replications.
Y_sim: Simulated final BSTS-based forecasts.
MSE: Mean squared error for each simulation run.

Arguments

nsim: Integer, number of simulation replications.
n_train: Integer, number of training observations.
n_test: Integer, number of test observations.
copula: Character, specifying the copula type: "Clayton", "Frank", "Gumbel", "Joe", or "Gaussian".
init_params: Numeric vector, initial parameter values for optimization.
fn: Function, log-likelihood function for parameter estimation.
U_train: Numeric matrix (n_train x D), pseudo-observations for the copula.
Z_train: Numeric array (n_train x D x M), observed data for each margin and sub-feature.
Z_test: Numeric array (n_test x D x M), observed data for each margin and sub-feature.
X_train: Numeric matrix (n_train x M), risk factors for the dynamic copula parameter.
X_test: Numeric matrix (n_test x M), risk factors for the dynamic copula parameter.
Y_test: Numeric matrix (n_test x D), true future values for MSE calculation.
BSTS_list: List of length D, each element is a BSTS model for a different margin.