harris-chris/joint-shapley-values

Source code for the Joint Shapley values: a measure of joint feature importance

joint-shapley-values's Issues

Joint Shap for DeepExplainer

Hi,

I am trying to adapt your "boston-housing-walkthrough" example to LSTM models with DeepExplainer.
I got an error below:
coalitions_to_k = get_powerset_to_k(n_features, 2)
explainer = shap.DeepExplainer(model, x_train)

baseline = explainer(x_test)

def value_function(cln_X):
return model.predict(cln_X)# - baseline[0]

TypeError Traceback (most recent call last)
in
2 explainer = shap.DeepExplainer(model, x_train)
3
----> 4 baseline = explainer(x_test)
5
6 def value_function(cln_X):

TypeError: 'DeepExplainer' object is not callable

Could you please help me? I could not find any example about joint shap with deep learning.