spikeSlabGAM(formula, data, ..., family="gaussian", hyperparameters=list(),
model=list(), mcmc=list(), start=list())fct),
linear/polynomial terms (lin), uni- or bivariate splines (sm, srf), random intercepts (rnd)
or Markov random fields (mrf) and their interactions,
i.e. an interaction between two smooth terms yields an effect surface, an interaction between a linear term and a random intercept yields random slopes,
an interaction between a linear term and a smooth term yields a varying coefficient term etc.
Implemented types of terms: [object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Terms in the formula that are not in the list of specials (i.e. the list of term types above) are automatically assigned
an appropriate special wrapper, i.e. numerical covariates x are treated as lin(x) + sm(x), factors f
(and numerical covariates with very few distinct values, see ssGAMDesign) are treated as fct(f).
Valid model formulas have to satisfy the following conditions:
y ~ x1 + x1:x2x2 is missing.
u) and penalized terms are not allowed, i.e.
y ~ u(x1)*x2 will produce an error.
y ~ lin(x1) + lin(x2) + x1:x2 will produce an error.spikeSlabGAM with methods
summary.spikeSlabGAM, predict.spikeSlabGAM, and plot.spikeSlabGAM.ssGAMDesign for details on model specification, spikeAndSlab for more details on the MCMC sampler and prior specification,
and ssGAM2Bugs for MCMC diagnostics. Check out the vignette for theoretical background and code examples.ssGAMDesign).}
ssGAMDesign}
"poisson" and "binomial" are implemented as well.}
spikeAndSlab.}
spikeAndSlab.
User-supplied groupIndicators and H entries will be overwritten by ssGAMDesign.}
spikeAndSlab.}
spikeAndSlab.}