Hi Alexandr,
On 11 November 2014 00:10, Alexandr M <rus...@gmail.com> wrote:
> Sorry that I formulated my question not very accurately.
> I form expressions/(logic conditions for parameters evidence and event)
> dynamically inside the loop and they are sometimes quite long.
By any chance, are using that for prediction? Because predict(...,
method = "bayes-lw") does posterior predictions from any set of
variables.
As an alternative, you can do cpdist(..., method = "lw") which also
generates from the posterior distribution:
> str(cpdist(fit, node = "A", evidence = list(B = "b"), method = "lw"))
Classes ‘bn.cpdist’ and 'data.frame': 10000 obs. of 1 variable:
$ A: Factor w/ 3 levels "a","b","c": 1 2 2 3 1 1 1 2 1 1 ...
- attr(*, "weights")= num 0.114 1 1 0.428 0.114 ...
- attr(*, "method")= chr "lw"
To compute P(A = a | whatever you conditioned on), just sum the
corresponding weights over the total weight mass. Since no language
trickery is involved, this works reliably.
Cheers,
Marco
--
Marco Scutari, Ph.D.
Lecturer in Statistics, Department of Statistics
University of Oxford, United Kingdom
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
