Variational Approximations for Generalized Additive Models
Simulate example datasets from a generalized additive models (GAM).
Basic plots for a fitted generalized additive model (GAMs).
Predictions from a fitted generalized additive model (GAM).
Summary of generalized additive model (GAM) fitted using variational a...
Variational approximations for generalized additive models
Fitting generalized additive models (GAMs) using variational approxima...
Fits generalized additive models (GAMs) using a variational approximations (VA) framework. In brief, the VA framework provides a fully or at least closed to fully tractable lower bound approximation to the marginal likelihood of a GAM when it is parameterized as a mixed model (using penalized splines, say). In doing so, the VA framework aims offers both the stability and natural inference tools available in the mixed model approach to GAMs, while achieving computation times comparable to that of using the penalized likelihood approach to GAMs. See Hui et al. (2018) <doi:10.1080/01621459.2018.1518235>.