GBMDiscretizer fails when estimator is XGBClassifier

[kyle-clara]

I get the following error with xgboost==1.6.2 and scikit-gbm==0.2.1:

    109 def xgboost_trees_to_dataframe(obj):
--> 110     trees_df = obj.get_booster().trees_to_dataframe()
    112     # Rename columns
    113     trees_df = \
    114         trees_df.rename(columns={
    115             'Tree'     : 'tree_index',
   (...)
    124             'Category' : 'category' 
    125         })

AttributeError: '_XgboostWrapper' object has no attribute 'get_booster'

Reproducible example:

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
import numpy as np

iris = load_iris()
data = pd.DataFrame(
    data= np.c_[iris['data'], iris['target']],
    columns= iris['feature_names'] + ['target']
)
data.columns = data.columns.str[:-5]
data.columns = data.columns.str.replace(' ', '_')

X, y = data.iloc[:, :4], data.iloc[:, 4:]
X_train, X_test, y_train, y_test = \
    train_test_split(X, y, test_size=0.3, random_state=0)
X_cols = X.columns.tolist()

gbm_discretizer = GBMDiscretizer(XGBClassifier(verbose=-1), X_cols, one_hot=False)
gbm_discretizer.fit_transform(X_train, y_train)

[madMathematician971]

I am facing the same issue as @kyle-clara. Also, if I try to use GradientBoostingRegressor\Classifier instead of XGBoost I get a different error:

I believe this library was written for an earlier version of XGBoost and sklearn as I found multiple posts on StackOverflow mentioning same errors with get_booster attr missing which were resolved by downgrading xgboost.

@krzjoa Could you please specify versions of sklearn and xgboost libraries this package was written for and (preferably) update the code to work with the latest versions? Thank you

[madMathematician971]

@kyle-clara My fix in #18 is verified, you can install the forked version from git until the author merges my MR