bac22734fb
- Remove `ntree_limit`. The parameter has been deprecated since 1.4.0. - The SHAP package compatibility is broken.
33 lines
832 B
Python
33 lines
832 B
Python
import numpy as np
|
|
import pytest
|
|
|
|
import xgboost as xgb
|
|
|
|
try:
|
|
import shap
|
|
except Exception:
|
|
shap = None
|
|
pass
|
|
|
|
|
|
pytestmark = pytest.mark.skipif(shap is None, reason="Requires shap package")
|
|
|
|
|
|
# xgboost removed ntree_limit in 2.0, which breaks the SHAP package.
|
|
@pytest.mark.xfail
|
|
def test_with_shap() -> None:
|
|
from sklearn.datasets import fetch_california_housing
|
|
|
|
X, y = fetch_california_housing(return_X_y=True)
|
|
dtrain = xgb.DMatrix(X, label=y)
|
|
model = xgb.train({"learning_rate": 0.01}, dtrain, 10)
|
|
explainer = shap.TreeExplainer(model)
|
|
shap_values = explainer.shap_values(X)
|
|
margin = model.predict(dtrain, output_margin=True)
|
|
assert np.allclose(
|
|
np.sum(shap_values, axis=len(shap_values.shape) - 1),
|
|
margin - explainer.expected_value,
|
|
1e-3,
|
|
1e-3,
|
|
)
|