In [30]:

    

%run find_signature.py


    

In [31]:

    

len(features_train)


    

    
Out[31]:





150

In [32]:

    

clf.score(features_test, labels_test)


    

    
Out[32]:





0.81683731513083047

In [33]:

    

importances = clf.feature_importances_


    

In [41]:

    

max_ = [i for i in importances if i > 0.2]


    

In [42]:

    

max_


    

    
Out[42]:





[0.36363636363636365]

In [45]:

    

a = list(importances)
idx = a.index(max_)


    

In [47]:

    

vectorizer.get_feature_names()[idx]


    

    
Out[47]:





u'houectect'

In [ ]: