如何从 H2O 随机森林中获得一棵树？

Question

I'm doing a random forest with the next code:

rf_md <- h2o.randomForest(training_frame = train_h,
                          nfolds = 5,
                          y = y,
                          ntrees = 500,
                          stopping_rounds = 10,
                          stopping_metric = "RMSE",
                          score_each_iteration = TRUE,
                          stopping_tolerance = 0.0001,
                          seed = 1234)

The output is an object with type: h2oregressionmodel . I want to get a single tree from the forest, so I found out that the function getTree() works out very well with randomForest objects. How can I convert my actual output to a randomForest object? or Can I get the same random forest with a different function which its output is a randomForest object?

Appreciate your help

Answer 1

H2O comes with a similar function, h2o.getModelTree , which can be used for both GBM and Random Forest models (see the docs ); in your case, for selecting, say, tree #3, it should be:

tree <- h2o.getModelTree(model=rf_md, tree_number=3)

Answer 2

In your example above, rf_md is the H2O-3 DRF (Distributed Random Forest) object. (Or, more specifically, a pointer to it. The real object lives in memory in the H2O-3 java process from the h2o.init())

H2O-3 model objects are different from the native R package objects. There is no way to convert an H2O-3 DRF model into an R randomForest model.