bamlss.engine.setup.RdThis function takes the x object of a bamlss.frame and adds
additional objects that are useful for model fitting engines. This is applied only
for 'regular' terms, e.g., as created by s and te.
For special model terms the corresponding smooth.construct method is in charge
of this (see also the examples for function bfit).
bamlss.engine.setup(x, update = "iwls", propose = "iwlsC_gp",
do.optim = NULL, df = NULL, parametric2smooth = TRUE, ...)The x list, as returned from function bamlss.frame,
holding all model matrices and other information that is used for
fitting the model.
Sets the updating function for model terms, see function bfit.
Sets the propose function for model terms, see function GMCMC.
Adds list element "do.optim" in the "state" element, see the
details below.
The initial degrees of freedom that should be assigned to a smooth model term,
based on the trace of the smoother matrix. Note that df can be a named numeric vector.
If the names match the labels of the model terms, the corresponding df are used, e.g.,
df = c("s(x1)" = 1, "s(x2)" = 2) sets different dfs for each term.
Should parametric model terms be transformed into an artificial
smooth model term and be added to the "smooth.construct" object within the x list?
This feature is handy, since algorithms can then cycle over the "smooth.construct"
object, only.
Currently not used.
For each model term in the "smooth.construct" object of the x list (as returned
from bamlss.frame), this function adds a named list called "state" with the
following entries:
"parameters": A numeric vector. Regression coefficients are named with "b",
smooth variances are named with "tau2".
"fitted.values": Given the "parameters", the actual fitted values of
the model term.
"edf": Given the smoothing variances, the actual equivalent degrees of freedom
(edf) of the model term.
"do.optim": Should an optimizer function try to find optimum smoothing variances?
The state will be changed in each iteration and can be passed outside an updating function.
Additionally, if missing in the xt argument of a model term (see, e.g., function
s for xt) the function adds the corresponding log-prior and its first
and second order derivatives w.r.t. regression coefficients in functions grad() and
hess().
Also, objects named "lower" and "upper" are added to each model term. These
indicate the lower and upper boundaries of the parameter space.
A transformed x list, as returned from function bamlss.frame.
d <- GAMart()
bf <- bamlss.frame(num ~ s(x1) + s(x2), data = d, family = "gaussian")
names(bf$x$mu$smooth.construct)
#> [1] "s(x1)" "s(x2)"
bf$x <- bamlss.engine.setup(bf$x, df = c("s(x1)" = 1, "s(x2)" = 3))
names(bf$x$mu$smooth.construct)
#> [1] "s(x1)" "s(x2)" "model.matrix"
names(bf$x$mu$smooth.construct[["s(x1)"]])
#> [1] "term" "bs.dim" "fixed" "dim"
#> [5] "p.order" "by" "label" "xt"
#> [9] "id" "sp" "drop.null" "X"
#> [13] "S" "UZ" "Xu" "df"
#> [17] "shift" "rank" "null.space.dim" "plot.me"
#> [21] "side.constrain" "repara" "S.scale" "X.dim"
#> [25] "fit.fun" "prior" "grad" "hess"
#> [29] "binning" "nobs" "weights" "rres"
#> [33] "fit.reduced" "fxsp" "a" "b"
#> [37] "edf" "lower" "upper" "state"
#> [41] "sparse.setup" "added" "pid" "update"
#> [45] "propose"
names(bf$x$mu$smooth.construct[["s(x1)"]]$state)
#> [1] "parameters" "fitted.values" "edf" "do.optim"
sapply(bf$x$mu$smooth.construct, function(x) {
c(x$state$edf, get.state(x, "tau2"))
})
#> $`s(x1)`
#> tau21
#> 1.000000e+00 2.591232e-05
#>
#> $`s(x2)`
#> tau21
#> 3.0000000 0.5346005
#>
#> $model.matrix
#>
#> 1
#>