Is there are some ways to know on which subsample of data the XGBoost tree was fitted?

Question

I'm making some XGBoost practice and I'd like to know on which subset of data the trees of XGBRegressor was fitted. Here the list of params that I use:

params = {'learning_rate': 0.09, 
          'n_estimators': 5,
          'objective': 'reg:squarederror',
          'max_depth': 6,
          'subsample': 0.2 
         }

I'm working with Python 3.7 and XGBoost library. So is there some way to know that? I've read the doc but couldn't find some useful.

score 0 · Answer 1 · answered Feb 11 '20 at 14:08

0

you can easily draw the tree and check it out you can do it with

export_graphiviz(uralgo,"filename",feature_names,rounded=True)

answered Feb 11 '20 at 14:08

parsa

370
4
13

it shows only a tree, but I need an indexes or rows of my dataset that were used to fit that tree – Kyrylo Chernov Feb 11 '20 at 14:25

Is there are some ways to know on which subsample of data the XGBoost tree was fitted?

1 Answers1