1

I'm using the package LIME in Python 3.7 but the output doesn't seem to be intuitive in some variables.

def explain(index, num_features):
    exp = explainer.explain_instance(X_test.loc[index,:], rf.predict, num_features = num_features)
    exp.show_in_notebook(show_all=False) #only the features used in the explanation are displayed
    print(exp.as_list())

num_feat = 5
explain(5000, num_feat)

I get the following results which work as it should

LIME output

However, if you look at the feature value for LandUse6 the variable importance from LIME is stricly larger than 0 and the feature value is 0. Is this not fulfilled or how do I interpret this value?

Sebastian Balle
  • 61
  • 7

0 Answers0