How to extract support vectors from SVM classifier in Python?

Question

I'm using LinearSVM from SciKit library in python to classify text data, and it works perfectly. My question is that is there a way to find the support vectors of my classifier model? I think the support vectors for my data would be a list of words!

I need this because I want to find a distinguishing feature of the classifier for different text data. (How different text data are distinct)

Thanks in advance.

Answer 1

For the SVM case in scikit-learn you should be able to access the support vectors in the following way:

>>> # get support vectors
>>> clf.support_vectors_
array([[ 0.,  0.],
       [ 1.,  1.]])
>>> # get indices of support vectors
>>> clf.support_ 
array([0, 1]...)
>>> # get number of support vectors for each class
>>> clf.n_support_ 
array([1, 1]...)

[Source: http://scikit-learn.org/stable/modules/svm.html ]