| cut {INLA} | R Documentation |
Group-wise model criticism using node-splitting
Description
This function performs group-wise, cross-validatory model assessment for an
INLA model using so-called node-splitting (Marshall and Spiegelhalter, 2007;
Presanis et al, 2013). The user inputs an object of class inla (i.e.
a result of a call to inla()) as well as a variable name
(split.by) specifying a grouping: Data points that share the same
value of split.by are in the same group. The function then checks
whether each group is an "outlier", or in conflict with the remaining
groups, using the methodology described in Ferkingstad et al (2017). The
result is a vector containing a p-value for each group, corresponding to a
test for each group i, where the null hypothesis is that group
i is consistent with the other groups except i (so a small
p-value is evidence that the group is an "outlier"). See Ferkingstad et al
(2017) for further details.
Usage
inla.cut(result, split.by, mc.cores = NULL, debug = FALSE)
Arguments
result |
An object of class |
split.by |
The name of the variable to group by. Data points that have
the same value of |
mc.cores |
The number of cores to use in |
debug |
Print debugging information if |
Value
A numeric vector of p-values, corresponding to a test for each group
i where the null hypothesis is that group i is consistent with
the other groups except i. A small p-value for a group indicates that
the group is an "outlier" (in conflict with remaining groups).
This function is EXPERIMENTAL!!!
Author(s)
Egil Ferkingstad egil.ferkingstad@gmail.com and Havard Rue hrue@r-inla.org
References
Ferkingstad, E., Held, L. and Rue, H. (2017). Fast and accurate Bayesian model criticism and conflict diagnostics using R-INLA. arXiv preprint arXiv:1708.03272, available at http://arxiv.org/abs/1708.03272. Published in Stat, 6:331-344 (2017).
Marshall, E. C. and Spiegelhalter, D. J. (2007). Identifying outliers in Bayesian hierarchical models: a simulation-based approach. Bayesian Analysis, 2(2):409-444.
Presanis, A. M., Ohlssen, D., Spiegelhalter, D. J., De Angelis, D., et al. (2013). Conflict diagnostics in directed acyclic graphs, with applications in Bayesian evidence synthesis. Statistical Science, 28(3):376-397.
Examples
## See http://www.r-inla.org/examples/case-studies/ferkingstad-2017 and Ferkingstad et al (2017).