I'm using the h2o package in Python.
When binomial_double_trees == True, I want to know information about internal all trees for a class.
This is my parameter:
h2o_rfe = H2ORandomForestEstimator(ntrees=3, max_depth=12, min_rows=10, binomial_double_trees=True)
If I use:
h2o_rfe.predict_leaf_node_assignment()
I can see that each tree is classified according to a separate tree and is also in h2o flow web page.
However, if I use:
list_of_trees = [H2OTree(model=h2o_rfe, tree_number=t, tree_class=None) for t in range(h2o_rfe.params['ntrees']['actual'])]'
I can't know the information of each internal tree according to the class.
Also I tried to set the tree_class (domain is 0,1):
list_of_trees = [H2OTree(model=h2o_rfe, tree_number=t, tree_class="0") for t in range(h2o_rfe.params['ntrees']['actual'])]'
When tree_class is '0', it is okay:
list_of_trees = [H2OTree(model=h2o_rfe, tree_number=t, tree_class="1") for t in range(h2o_rfe.params['ntrees']['actual'])]'
But when tree_class is '1', this error occurs:
Error: For binomial, only one tree class has been built per each iteration: 0
How to check the internal trees?
