i have text data and pip line model . i want to using shap library to Visualize the impact on all the output classes
i got this error :
TypeError: The passed model is not callable and cannot be analyzed directly with the given masker! Model: Pipeline(steps=[('vect1', CountVectorizer()),
('tfidf1', TfidfTransformer()), ('clf1', MultinomialNB())])
how can i handle it
try this :
vectorized_data = pipe_model1.named_steps['vect1'].transform(background_data)
transformed_background_data =pipe_model1.named_steps['tfidf1'].transform(vectorized_data)
explainer = shap.KernelExplainer(pipe_model1.named_steps['clf1'].predict_proba, transformed_background_data)
shap_values = explainer.shap_values(transformed_background_data)
the background_data is your data to use
your using the pipeline model and you should add your pip line model sequensly
