Is it possible to use SHAP with sklearn pipeline?

[santokalayil]

Is it possible to use SHAP with sklearn pipeline (in my use case imblearn pipeline) where final estimator is XGBoost?

[santokalayil]

I need to use this pipe:

transformers_list.append(("category", OneHotEncoder(handle_unknown="ignore"), features_dict['cat_features']))
transformers_list.append(("numeric", RobustScaler(), features_dict['num_features']))

preprocessor = ColumnTransformer(
    transformers=transformers_list,
    remainder="passthrough", # error(default)
)
pipe = Pipeline(steps=[
    ("preprocessor", preprocessor), 
    ("over_sampling", SMOTE()), 
    ("classifier", clf_xgb)
] )```



with something like this:

model = BoostBoruta(
pipe._final_estimator, param_grid=param_grid, max_iter=200, perc=100,
importance_type='shap_importances', train_importance=False
)```

[cerlymarco]

@santokalayil all classes available in shap-hypetune are developed as sklearn BaseEstimator. This means that you can simply use them as any other sklearn models.

If you support the project don't forget to leave a star ;-)

all the best