Tools

Functions:

trees_to_dataframe(obj)

A common interface to fetch trees as pandas DataFrame
skgbm.tools.trees_to_dataframe(obj)[source]

A common interface to fetch trees as pandas DataFrame

DaraFrame columns for all the models share their names, but they differ when it comes to the exact set of available parameters.

XGBoost

LightGBM

CatBoost

scikit-learn

tree_index

✅

✅

✅

✅

node_depth

❌

✅

✅

✅

node_index

✅

✅

✅

❌

left_child

✅

✅

✅

✅

right_child

✅

✅

✅

✅

parent_index

❌

✅

❌

❌

split_feature

✅

✅

✅

✅

split_gain

✅

✅

❌

❌

threshold

✅

✅

✅

✅

decision_type

❌

✅

❌

✅

missing

✅

✅

❌

❌

missing_type

❌

✅

❌

❌

value

❌

✅

✅

✅

weight

❌

✅

✅

❌

count

✅

✅

❌

✅
Parameters:

obj (object) – An XGBoost, LightGBM, CatBoost or scikit-learn GradientBoosting* model

Returns:

trees_df – A pandas DataFrame containing information about all the trees in the model

Return type:

pd.DataFrame

Examples

>>> from skgbm.tools import trees_to_dataframe
>>> from sklearn.datasets import make_regression
>>> from sklearn.ensemble import GradientBoostingRegressor
>>> X, y = make_regression()
>>> gb_reg = GradientBoostingRegressor().fit(X, y)
>>> gb_df = trees_to_dataframe(gb_reg)