SHAP分析lightGBM_lightgbm shap

SHAP (SHapley Additive exPlanations) is a game theoretic approach to explain the output of any machine learning model. It connects optimal credit allocation with local explanations using the classic Shapley values from game theory and their related extensions (see papers for details and citations).

git 地址：https://github.com/slundberg/shap

代码：
valid_df = pd.read_pickle(f’…/input/riiid-cv2/cv1_valid.pickle’)
X_valid=valid_df[features]
X_valid=X_valid[0:1000]
model = lgb.Booster(model_file=f’…/input/riiid-lgbm-bagging-train2/model1.txt’)

model.params[“objective”] = “binary”
explainer = shap.TreeExplainer(model)
shap_values = explainer.shap_values(X_valid)
shap.summary_plot(shap_values, X_valid,max_display=50) #特征重要性可视化