How can I manually add a root node in XGBoost tree(s)?

Question

I have a binary feature {0, 1} and only when its value is 1, I would like my XGBoost model to "evaluate" a set of trees. The goal is to save on prediction time by not evaluating a set of trees when the binary feature's value is 0.

So I was thinking if this feature is the root node of those trees, with rest of the tree being the subtree on the value of 1 of this feature, I may achieve my goal.

Is it possible to tweak XGBoost model in such a way?

Answer 1

It would be best to add that as control logic outside of the model.

If you were using Python, it would be something like:

from xgboost import XGBClassifier

model = XGBClassifier()
model.fit(X_train, y_train)

if feature == 1:
    y = model.predict(X)
else feature == 0:
    print("No prediciton made.")