This function implements the variable selection for Gaussian process-based SVC models using a penalized maximum likelihood estimation (PMLE, Dambon et al., 2021, <arXiv:2101.01932>). It jointly selects the fixed and random effects of GP-based SVC models.
SVC_selection(obj.fun, mle.par, control, ...)(SVC_obj_fun)
Function of class SVC_obj_fun. This is the output of
SVC_mle with the SVC_mle_control parameter
extract_fun set to TRUE. This objective function comprises
of the whole SVC model on which the selection should be applied.
(numeric(2*q+1))
Numeric vector with estimated covariance parameters of unpenalized MLE.
(list)
List of control parameters for variable selection. Output of
SVC_selection_control.
Further arguments.
Returns an object of class SVC_selection. It contains parameter estimates under PMLE and the optimization as well as choice of the shrinkage parameters.
Dambon, J. A., Sigrist, F., Furrer, R. (2021). Joint Variable Selection of both Fixed and Random Effects for Gaussian Process-based Spatially Varying Coefficient Models, ArXiv Preprint https://arxiv.org/abs/2101.01932