stages_simplebhc: Backward hill-climbing for simple staged trees
Description
Greedy search of simple staged event trees
with iterative joining of positions.
Usage
stages_simplebhc(
object,
score = function(x) {
return(-BIC(x))
},
scope = NULL,
max_iter = Inf,
ignore = object$name_unobserved
)Value
an object of class sevt, the simple staged tree resulting
from the search.
Arguments
- object
an object of class sevt
with fitted probabilities and data,
as returned by full or
sevt_fit.
- score
the score function to be maximized.
- scope
names of variables that should be considered for
the optimization.
- max_iter
the maximum number of iterations per variable.
- ignore
vector of stages which will be ignored and left untouched,
by default the name of the unobserved stages stored in
object$name_unobserved.
Details
This function is similar to the classical
backward hill-climbing implemented in stages_bhc, but
instead of joining stages it consider joining of positions via
join_positions.
Thus, the search is in the space of simple staged tree models if the
initial stage tree is simple.
See the references for additional details.
References
Leonelli M, Varando G.
Structural Learning of Simple Staged Trees,
arXiv preprint
arXiv:2203.04390v1
See Also
join_positions()
sevt_simplify()
Examples
Run this codemod <- stages_simplebhc(full(Titanic))
plot(mod)
Run the code above in your browser using DataLab