logbin.smooth,
logbin.smooth.allref takes the formula and data
for a log-link binomial GLM with smooth terms and produces a
list of all parameterisations needed for the CEM algorithm
associated with the semi-parametric part of the model.
logbin.smooth.allref(object, data = environment(object), mono, logbin.smooth.spec, num.knots)terms object for the "fake.formula"
associated with a logbin.smooth model
(see interpret.logbin.smooth).
get_all_vars for
the fake.formula.
fake.formula should be restricted to have a
monotonically non-decreasing relationship with the
outcome. May be specified as names or indices of the
terms.
interpret.logbin.smooth.
NA for Iso terms).
terms component of object.data argument.terms are restricted to be
monotonically non-decreasing.logbin.smooth use an extended CEM algorithm
by partioning the parameter space associated with the smooth terms into a
collection of restricted parameter spaces, each corresponding to a
restricted fully parametric model that can be fit using logbin.
This is a workhorse function that creates the list of possible parameterisations of
each smooth term.Isotonic terms and monotonic B-spline terms have only one parameterisation: where the maximum fitted value occurs at the maximum of the covariate range.
Unrestricted B-spline terms each have $k + 3$ parameterisations (where $k$ is the number of internal knots), corresponding to the possible locations of the maximum of the smooth curve along the range of the covariate.
logbin.smooth considers all possible combinations of the number of
knots for each smooth term, and all possible combinations of the associated
parameterisations, and logbin.smooth.design creates the appropriate
formula and design matrix to be used in the call to logbin.
logbin.smooth