marjoleinF/gamtree: GAM-based Recursive Partitioning

Recursively partitions the observations in a dataset, based on the parameters of a generalized additive model (GAM). Function splinetree implements partitioning based on parametric or non-penalized spline models. It offers faster computational speed and more flexibility for modeling multilevel and longitudinal data. Function gamtree implements partitioning based on semi-parametric or penalized spline models, a.k.a. smoothing splines. It requires less user involvement in correctly specifying the splines and knots, but has a heavier computational load and offers less flexibility in modeling multilevel and longitudinal data. Packages mgcv and splines are used for spline and model estimation, packages partykit and merDeriv are used for partitioning and derivative computations.

install.packages("remotes")
remotes::install_github("marjoleinF/gamtree")