Can we use pre-trained Decision Tree or Random Forest model as base model for XGBoost?

#1

Is it possible to use pre-trained Decision Tree or Random Forest model with XGBoost algorithm ?

If yes then which parameter of XGBoost do we use to specify the base model (which could be a pre-trained Decision Tree or Random Forest)?

I’m trying to work with XGBoost Regressor for a regression problem.

#2

No, it is not possible to do so. You may however specify the base_margin array in fit() to specify the initial set of predictions from which boosting should start:

clf = xgb.XGBRegressor(...)
clf.fit(X, y, base_margin=base_margin)