restoreParams: Reparametrise bugs output

In glmmBUGS: Generalised Linear Mixed Models with BUGS and JAGS

Description

Undoes the parametrisation used in writeBugsModel, and gives the original names to random effect levels.

Usage

restoreParams(bugsResult, ragged = NULL,extraX=NULL)

Arguments

bugsResult	Output from bugs, using a ragged array generated by winBugsRaggedArray and a model generated by writeBugsModel
ragged	The ragged array used to call bugs
extraX	Possible extra covariates for spatial regions with no data but do have predicted spatial effects.

Value

A list where each element is a matrix or an array. The first dimension is the number of realisations, the second the number of chains, and for vector-valued parameters and random effects, the third dimension is the length of the parameter.

If the model contains a spatial component, the result will have list entries the following:

Reffect

The random effect. In the case of spatial models this is the sum of the spatial and non-spatial random effects U+V.

,

ReffectSpatial	The spatial random effect for each region, if any
FittedReffect	The predicted values on the link scale, being the random effect plus intercept and effect of covariates.

Note

For spatial models, one fitted rate is computed for each region in the adjacency matrix, even though some of these regions may not have spatial or non-spatial random effects simulated in the bugs model. If a spatial random effect is missing (as happens with islands), a zero is added. If a non-spatial random effect is missing (as happens when a regions does not have data), a value is simulated unconditionally from each iteration's intercept and standard deviation for that effect. Note that this does not add on the effect of possible covariates for that region. This can be added via the extraX argument.

Author(s)

Patrick Brown [e-mail]

See Also

bugs

glmmBUGS documentation built on May 2, 2019, 6:08 p.m.