Hi all
I have a problem with the cpquery function in the bnlearn package.
I have constructed a hybrid network (using a mix of continuous and discrete
variables).
The network is named "fitted".
I am interested in predicting the probability of observing a value greater
that a particular threshold in a continuous variable (ABW) from evidence
presented in a mix of continuous and discrete variables.
One of the continuous variables (EST) takes 3 values, x, y and z.
I find that the cpquery gives very plausible results if the EST is set to
equal x.
For example:
> cpquery(fitted,event=(ABW>=11), evidence=eval(parse(text="(EST=='x' & TR>9
& BU>15819 & RF>2989)")),n=10^6)
[1] 0.7471088
While inverting the threshold for ABW gives:
> cpquery(fitted,event=(ABW<=11), evidence=eval(parse(text="(EST=='x' & TR>9
& BU>15819 & RF>2989)")),n=10^6)
[1] 0.2587795
However, if I replace EST=='x' with EST=='z' or EST=='y' I get 0 probability
of obtaining a value for ABW that is either greater or less than the
threshold.
For example:
> cpquery(fitted,event=(ABW>=11), evidence=eval(parse(text="(EST=='y' & TR>9
& BU>15819 & RF>2989)")),n=10^6)
[1] 0
and
> cpquery(fitted,event=(ABW<=11), evidence=eval(parse(text="(EST=='y' & TR>9
& BU>15819 & RF>2989)")),n=10^6)
[1] 0
I do not understand why classes y and z are returning 0 probabilities for
these queries.
My own knowledge from the data is that these classes should both typically
return a value for ABW that is very much higher than the threshold value.
Can you help me understand why my cpquery code is not giving the anticipated
results for x and y?
Many thanks
Ross
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
