Smoothing parameter estimation methods mgcv offers

GCV.Cp is mgcv's default and what the package has always used. REML and ML are the ones to reach for when a smooth looks overfitted: generalized cross-validation is known to undersmooth, and REML resists it.

Usage

gam_methods()

Value

character vector of method names

References

Wood SN (2011). Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society: Series B 73(1), 3-36. doi:10.1111/j.1467-9868.2010.00749.x

Examples

gam_methods()
#> [1] "GCV.Cp" "REML"   "ML"     "P-REML" "P-ML"