Theory and Algorithms for Bayesian Networks 105
> gs(cropdata1, test = "cor", alpha = 0.05, debug = TRUE)
[...]
----------------------------------------------------------------
* learning the markov blanket of N .
[...]
* checking node V for inclusion.
> N indep. V given ' C ' (p-value: 0.3048644).
* checking node W for inclusion.
> node W included in the markov blanket (p-value: 7.89e-07).
> markov blanket ( 2 nodes ) now is ' C W '.
> restarting grow loop.
* checking node V for inclusion.
> N indep. V given ' C W ' (p-value: 0.1416876).
* checking node W for exclusion (shrinking phase).
> node W remains in the markov blanket. (p-value: 7.89e-07)
[...]
When learning the Markov blanket of N, we have that node C is included
before V. Apparently, the dependence ...