abn (version 0.8)

calc.node.mlik.inla: internal function for individual node computation via INLA

Description

This functions is for internal use only and not meant to be called by the user.

Usage

calc.node.mlik.inla(child.loc,dag.m.loc,data.df.loc,data.dists.loc,ntrials.loc,exposure.loc,compute.fixed.loc,
                         mean.intercept.loc,prec.intercept.loc,mean.loc,prec.loc,loggam.shape.loc,loggam.inv.scale.loc,verbose.loc)

Arguments

child.loc

row index of current node in dag.m

dag.m.loc

a matrix defining the network structure, a directed acyclic graph (DAG), see details for format. Note that colnames and rownames must be set

data.df.loc

a data frame containing the data used for learning the network, binary variables must be declared as factors

data.dists.loc

a named list giving the distribution for each node in the network, see details

ntrials.loc

a numeric vector giving the total number of trials, only applicable if the data comprise one or more binary variables. see details

exposure.loc

a numeric vector, giving the unit of exposure, only applicable if the data comprise one or more poisson variables, see details

compute.fixed.loc

a logical flag, set to TRUE if also want the marginal posterior distributions for the model parameters as well as the marginal likelihood

mean.intercept.loc

the prior mean of the Gaussian intercept term for each node

prec.intercept.loc

the prior precision of the Gaussian intercept term for each node

mean.loc

the prior mean in the Gaussian prior for each covariate at each node

prec.loc

the prior precision in the Gaussian prior for each covariate at each node

loggam.shape.loc

the prior shape parameter in the Gamma prior for the precision parameters at each node

loggam.inv.scale.loc

the prior inverse scale (=1/scale) parameter in the Gamma prior for the precision parameters at each node

verbose.loc

if true then provides some additional output, in particular the call used to inla()

Value

• A named list. If compute.fixed=FALSE then each entry in the list is the log marginal likelihood for each node, else each entry in this list is itself a list comprising of all the different outputs provided by inla.

Details

This function is called in fitabn and buildscorecache and deals with the actual call to INLA which builds up an appropriate string and is then passed into eval(parse()). The arguments all have the .loc postix due to a scoping issue using eval(parse()) which does not seem to use the parent.frame/calling environment scope.